PDF Hosted at the Radboud Repository of the Radboud University Nijmegen

PDF hosted at the Radboud Repository of the Radboud University Nijmegen

The following full text is a publisher's version.

For additional information about this publication click this link. http://hdl.handle.net/2066/169056

Please be advised that this information was generated on 2017-12-07 and may be subject to change. Paranimfen Paranimfen Daphne Everaerd 6521 JT Nijmegen Daphne Everaerd 6521 JT Nijmegen Ellen van der Holst Ellen van Janneke van der Holst van Janneke Van den Havestraat 44 Havestraat den Van Ellen van der Holst Ellen van Janneke van der Holst van Janneke Van den Havestraat 44 Havestraat den Van Mind the step Mind the step Mind the step Mind the step aansluitende receptie. receptie. aansluitende deze plechtigheid en de deze [email protected] precies in de Aula van de van in de Aula precies aansluitende receptie. receptie. aansluitende U bent van harte welkom bij bij welkom harte U bent van deze plechtigheid en de deze [email protected] Comeniuslaan 2 te Nijmegen Comeniuslaan 2 te [email protected] precies in de Aula van de van in de Aula precies [email protected] Radboud Universiteit Nijmegen, Nijmegen, Universiteit Radboud U bent van harte welkom bij bij welkom harte U bent van [email protected] verdediging van mijn proefschrift van verdediging Comeniuslaan 2 te Nijmegen Comeniuslaan 2 te UITNODIGING [email protected] Voor het bijwonen van de openbare de openbare bijwonen van het Voor Radboud Universiteit Nijmegen, Nijmegen, Universiteit Radboud Op woensdag 5 april 2017 om 14.30u verdediging van mijn proefschrift van verdediging UITNODIGING Voor het bijwonen van de openbare de openbare bijwonen van het Voor Brain changes in motor performance in motor changes Brain in cerebral small vessel disease disease vessel small in cerebral Op woensdag 5 april 2017 om 14.30u Brain changes in motor performance in motor changes Brain in cerebral small vessel disease disease vessel small in cerebral

an der Holst Mind the step the step Mind Mind the step the step Mind Ellen (H.M.) v Ellen (H.M.) van der Holst Ellen (H.M.) van Brain changes in motor performance in motor changes Brain in cerebral small vessel disease small vessel in cerebral Brain changes in motor performance in motor changes Brain in cerebral small vessel disease small vessel in cerebral

THEMIND LONG-TERM THE STEP RISK IN CEREBRAL OF VASCULAR SMALL DISEASE VESSEL AND DISEASE EPILEPSYBrain AFTER changes STROKE in INmotor YOUNG performance ADULTS Ellen (H.M.) van der HolstRENATE M ARNTZ 266 THEMIND LONG-TERM THE STEP RISK IN CEREBRAL OF VASCULAR SMALL DISEASE VESSEL AND DISEASE EPILEPSY Brain AFTER changes STROKE in INmotor YOUNG performance ADULTS Ellen (H.M.) van der HolstRENATE M ARNTZ 266 ISBN 978-94-6284-098-0 ISBN 978-94-6284-098-0

Mind the step in cerebral small vessel disease Brain changes in motor performance

Ellen (H.M.) van der Holst Mind the step in cerebral small vessel disease

Brain changes in motor performance

Auteur: Ellen (H.M.) van der Holst Cover ontwerp: Hans (J.H.M.) van der Holst Drukwerk: ProefschriftMaken || www.proefschriftmaken.nl ISBN: 978-94-6284-098-0

The studies in this thesis were carried out at the Department of Neurology of the Donders Institute for Brain, Cognition and Behaviour, Centre for Neuroscience, Radboud university medical centre, Nijmegen, the Netherlands with financial support by a VIDI innovational grant from the Netherlands Organisation for Scientific Research (NWO, grant 016.126.351; prof.dr. FE de Leeuw).

No part of this thesis may be produced in any form or by any means without written permission of the author or the publisher holding the copyright of the published articles. Mind the step in cerebral small vessel disease Brain changes in motor performance

Proefschrift

ter verkrijging van de graad van doctor aan de Radboud Universiteit Nijmegen op gezag van de rector magnificus

prof. dr. J.H.J.M. van Krieken,

volgens besluit van het college van decanen in het openbaar te verdedigen op woensdag 5 april 2017 om 14.30 uur precies

door

Helena Maria (Ellen) van der Holst Geboren op 31 juli 1984 te Nijmegen Promotoren: Prof. dr. H.F. de Leeuw Prof. dr. C.J.M. Klijn

Copromotoren: Dr. A.M. Tuladhar Dr. E.J. van Dijk

Manuscriptcommissie: Prof. dr. ir. N. Karssemeijer (voorzitter) Dr. V.G.M. Weerdesteyn Prof. dr. W.M. van der Flier (VUmc) Voor mijn ouders

Table of contents

Part I Introduction 9 Chapter 1 General introduction, aims and outline 11

Part II Cerebral small vessel disease and cognitive performance 21 Chapter 2 Cingular integrity and verbal memory performance in cerebral small vessel disease 23

Part III Cerebral small vessel disease and motor performance 41 Chapter 3 Baseline cerebral small vessel disease and gait decline 43 Chapter 4 White matter changes and gait decline 61 Chapter 5 Baseline cerebral small vessel disease and incident parkinsonism 75

Part IV Long-term mortality in cerebral small vessel disease 91 Chapter 6 Factors associated with 8-year mortality in cerebral small vessel disease 93

Part V Summary and discussion 111 Chapter 7 General discussion and future perspectives 113 Chapter 8 Summary 129 Chapter 9 Summary in Dutch | Nederlandse samenvatting 135

Part VI Appendices 143 A1 List of abbreviations 145 A2 References 151 A3 Acknowledgements | Dankwoord 165 A4 Curriculum vitea 173 A5 List of publications 177 A6 Dissertations of the disorders of movement research group, Nijmegen 183 A7 Donders Graduate School for Cognitive Neuroscience Series 191

Part I Introduction

1. General introduction, aims and outline CHAPTER 1

12 GENERAL INTRODUCTION, AIMS AND OUTLINE.

Cerebral small vessel disease and imaging Cerebral small vessel disease (CSVD) is one of the most prevalent acquired vessel disorders in 1 the ageing human brain. CSVD encompasses degenerative alterations of various aetiologies 1 in the vessel wall of the small perforating cerebral arteries, arterioles, venules and capillaries.2 These vessel wall changes can lead to ischemic and/or haemorrhagic damage to the brain tissue supplied, including the white matter, brainstem, deep grey nuclei and the cortex. Main risk factors for this vessel wall damage are longstanding arterial hypertension, smoking and 2 diabetes.3, 4 Since these small vessels are difficult to image and to investigate in vivo, brain parenchymal lesions of presumed CSVD origin are adopted as imaging markers of CSVD.5 Common signs of CSVD on conventional magnetic resonance imaging (MRI) include: areas of incomplete infarction or chronic hypoperfusion (white matter hyperintensities (WMH)), small 3 areas of focal necrosis (lacunes), vessel wall rupture, manifesting by perivascular hemosiderin deposits (cerebral microbleeds) and brain atrophy (Box 1).2, 5 These signs on MRI are regarded as the traditional CSVD markers. International criteria on the terminology and definitions of these traditional CSVD markers have recently been published as the standards for reporting 4 vascular changes on neuroimaging (STRIVE) criteria.5 In the last decade, the imaging spectrum of CSVD has been extended to more subtle changes of the white matter, since the traditional MRI markers of CSVD mentioned above are regarded as the end of a continuous spectrum of white matter pathology. With diffusion 5 tensor imaging (DTI) the microstructural integrity of the white matter can be assessed. DTI is an MRI technique that provides quantitative measures of the mobility of water molecules in vivo (Box 2).6 DTI has shown to be sensitive to detect tissue damage, showing abnormalities in both WMH and in apparently normal appearing white matter on conventional MRI.7 This 6 imaging technique might capture CSVD-related lesions at earlier stages, presumably even before the traditional CSVD markers appear on conventional neuroimaging, since changes in normal appearing white matter showed tissue pathology less marked than those found in WMH.8 Although DTI has a low specificity in detecting the underlying cause of cerebral white 7 matter damage, DTI holds promise for giving further insight into the mechanisms underlying clinical symptoms of CSVD. The traditional CSVD markers are frequently reported on brain imaging of older adults 8 and their presence rises markedly with increasing age and cardiovascular risk factors. In population-based studies, WMH are found in >90% in individuals aged 60 years and over.9 Lacunes and microbleeds are less frequently seen; their presence varies considerable across studies and is related to the study population and the brain imaging protocol. Lacunes 9 were found in 11% in individuals aged 60-69 years10 and reaches to >30% in those aged 80 years and over in a population-based study.11 A similar pattern for microbleeds is seen; their prevalence is around 18% in individuals aged 60-69 years, up to 38% in those aged 80 years 12 and over in the general population. A In this thesis, the focus is on common sporadic CSVD, which is the most prevalent form of CSVD and includes age-related and cardiovascular risk-factor-related CSVD.2 Other forms

13 CHAPTER 1 of CSVD, including hereditary forms (e.g. cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL) or the recessive form (CARASIL) or Fabry disease), inflammatory and immunologically mediated CSVD (e.g. small vessel vasculitis) and venous collagenosis are left beyond the scope of this thesis.

Box 1: Conventional MRI characteristics of lesions related to cerebral small vessel disease Based on Wardlaw et al. Lancet Neurol 2013;12:822-8385

White matter hyperintensities of presumed vascular origin (figure A) White matter hyperintensities (WMH) are characterized by bilateral, typically symmetrical hyperintensities on T2-weighted MRI, including fluid-attenuated inversion recovery (FLAIR) scan. On T1-weighted sequences they can appear as isointense or hypointense. Their diameter is variable; they can range from small focal lesions to more confluent areas in the white matter.

Lacunes of presumed vascular origin (figure B) A lacune is defined as a round or ovoid, subcortical and fluid-filled cavity, with the same signal intensity as cerebrospinal fluid (CSF). The diameter is between 3-15 mm. They can be identified on FLAIR images, and have a central CSF-like hypointensity with usually a surrounding hyperintensive rim.

Cerebral microbleeds (figure C) Microbleeds are round or ovoid lesions that can be visualized on gradient-echo T2*- weighted imaging or susceptibility-weighted sequences as hypointense or black lesions with associated blooming, indicating hemosiderin deposits. They have generally a diameter between 2-5 mm, with a maximum of 10 mm.

14 GENERAL INTRODUCTION, AIMS AND OUTLINE.

Brain atrophy (figure D) Brain atrophy can occur in many disorders and has also been documented in cerebral small vessel disease (CSVD). Brain atrophy in CSVD is defined as a lower brain volume, not related 1 to macroscopic focal injury, as trauma or infarction. Tissue loss in the presence of substantial load of CSVD, including WMH seen on the FLAIR image in Figure D (left image), especially occurs by sulcal widening (arrow A) and ventricular enlargement (arrow B). 2 DD BaselineBaseline 2006 Follow-upFollow-up 2011

B 3

A BrainBrain atrophy A

15 CHAPTER 1

Box 2: diffusion tensor imaging Based on Alexander et al. Neurotherapeutics 2007;4:316-3296

Diffusion tensor imaging (DTI) is an MRI technique that measures the directionality and magnitude of random movement of water molecules in tissue. It is used to study the microstructural integrity of the white matter of the brain. Without barriers, water molecules move freely and uniformly in all directions in a spherically symmetric manner which is termed isotropic diffusion. In the presence of barriers, such as nerve fibres, the diffusion is not equal in all different directions, but is larger in one direction than in the other. This is called anisotropic diffusion.

A Isotropy Low anisotropy High anisotropy

A measure for the directionality of water is fractional anisotropy (FA) (Figure A). FA is zero for isotropic diffusion (i.e. the fraction of anisotropic diffusion is zero) and approaches the value of one as the diffusion becomes more and more ellipsoid in shape.

Fractional anisotropy

Diffusion of water at each voxel is characterized B by its three principal eigenvectors with associated l1 eigenvalues (λ1, λ2, λ3) (Figure B). l3 The average of the three eigenvalues represents the overall magnitude of water diffusion and is referred as the mean diffusivity (MD). It is expressed in mm2/s. l2 Axial diffusivity (AD) represents the magnitude of diffusivity parallel to the white matter tracts λ( 1). The Mean diffusivity: (l1 + l2 + l3 )/3 average of λ2 and λ3 is termed radial diffusivity (RD) Axial diffusivity: l1 and reflects the magnitude of diffusion perpendicular Radial diffusivity: (l2 + l3)/2 to these tracts.

In intact axons the diffusion of water will be mostly in one main direction, reflected by a high FA and the overall magnitude of diffusion will be restricted as water molecules have less space to move, resulting in a low MD. In general, a lower FA and a higher MD are associated with a poorer white matter microstructural integrity.13

16 GENERAL INTRODUCTION, AIMS AND OUTLINE.

Motor consequences of cerebral small vessel disease While CSVD has been regarded as an incidental finding with no clinical and therapeutic consequences for a long time, during the last two decades it is increasingly being recognized 1 as a serious problem, which causes major health problems in the aging society, including cognitive decline and dementia, depression and stroke.14 So far, the motor consequences of CSVD are relatively understudied. Gait is the resultant of the performance of many organ systems, including the peripheral and 2 central nerve system, cardiovascular and pulmonary system and musculoskeletal system.15 As a result, gait disturbances can have many causes. There is now emerging evidence (mainly at the cross-sectional level) that CSVD is one of most important vascular contributors to gait disturbances.16, 17 The mechanisms of how CSVD can lead to gait disturbances are 3 not entirely understood; it is thought that CSVD result in white matter tract disruption and loss of connectivity between brain areas. CSVD has also been associated with mild parkinsonian signs18 and especially with higher Unified Parkinson’s Disease Rating Scale (UPDRS) scores on gait, posture and postural stability.19 The association between CSVD and 4 parkinsonism is however controversial and evidence is mainly coming from post-mortem pathology studies. The presence of CSVD in patients with parkinsonism in the absence of the typical histopathology findings compatible with parkinsonism, including Lewy bodies or tau inclusions, is often referred as vascular parkinsonism.20, 21 It is unclear whether gait 5 disturbances and parkinsonism represent a clinical continuum in patients with CSVD, with isolated mild gait disturbances at the beginning and more severe gait disturbances and parkinsonism as the end-stage disease, or if they represent distinctive diseases with separate pathologies. Longitudinal studies investigating the role of CSVD in the development of 6 parkinsonism are currently lacking. There are several reasons why it is of interest to study the contribution of CSVD to gait disturbances and parkinsonism by using conventional MRI and DTI. First of all, it can provide insight into the pathophysiological mechanisms underlying these symptoms. Secondly, if 7 CSVD is proven to be associated with gait decline and parkinsonism, CSVD imaging markers might be used as surrogate markers for these motor symptoms in clinical trials, since clinical symptoms often develop in a late stage of the disease. Furthermore, it may open avenues 8 for possible interventions (e.g. therapeutic approaches or lifestyle changes) in an earlier disease stage of CSVD, thereby delaying or preventing gait disturbances and parkinsonism and minimizing their influence on the quality of life of older adults by reducing the risk of their adverse consequences, including falls, cognitive impairment, institutionalization and 9 death, which socially and economically burden the society.

17 CHAPTER 1

Mortality in cerebral small vessel disease As already outlined above, the long-term prognosis of CSVD is highly variable, and also death may be a consequence.22, 23 To date, it is not exactly known which patients with CSVD are at highest risk for these adverse outcomes, including mortality. Important clinical consequences of CSVD, including cognitive24 and gait disturbances25 have been associated with mortality in population-based studies. The place of neuroimaging in determining the risk of mortality in addition to these clinical parameters has never been investigated. Possibly, the association between these clinical parameters and mortality is driven by CSVD. Getting insight into determinants associated with mortality in an CSVD population might help to identify individuals at highest risk for experiencing these adverse events and provide information on factors reflecting the vital health status of individuals with CSVD.

Aim of this thesis and study design The aim of this thesis was to investigate the associations between imaging characteristics of CSVD and different clinical outcome measures, including cognitive disturbances, gait deterioration and the development of parkinsonism over time, in individuals with CSVD by using conventional MRI and DTI. In addition, we wanted to investigate which individuals with CSVD are at highest risk for an unfavourable outcome in the more distant future by identifying determinants, which best predicted 8-year mortality in these individuals. The studies presented in this thesis are based on the Radboud University Nijmegen Diffusion tensor imaging and Magnetic Resonance imaging Cohort (RUN DMC) study. This is a prospective cohort study that investigates the risk factors and clinical consequences of structural brain changes assessed by MRI among older adults with CSVD. The RUN DMC study compromised all consecutive patients referred to the department of Neurology of the Radboud University medical centre between October 2002 and November 2006, aged between 50-85 years with CSVD on brain imaging, defined as the presence of WMH and/or lacunes of presumed vascular origin. In total, 503 individuals with CSVD were included at the start of the study in 2006. Main exclusion criteria were parkinsonism or dementia at baseline, life-expectancy less than 6 months, non-CSVD related white matter lesions and MRI contra- indications. All participants underwent a brain MRI-scan and an extensive cognitive and motor assessment battery, including the assessment of gait and the presence of parkinsonian signs by using the motor section of the UPDRS (UPDRS-m). In 2011 this assessment, including a cerebral MRI was repeated, and all participants were screened for the presence of dementia and parkinsonism. The studies presented in this thesis are mainly based on the longitudinal data, except for chapter 2. For analysis on vital status, participants were followed until their death or until November 24, 2014.

18 GENERAL INTRODUCTION, AIMS AND OUTLINE.

Outline of the thesis Chapter 2 (part II) describes the cross-sectional association between microstructural integrity of the cingulum, an important white matter tract in cognitive performance, and 1 verbal memory performance. In part III, we explored the longitudinal association between imaging characteristics and motor performance after 5 years of follow-up. Chapter 3 reports on the association between baseline CSVD markers and gait decline after 5 years.Chapter 4 is an extension of the previous chapter and describes the longitudinal association between 2 progression of CSVD and loss of white matter integrity on follow-up imaging and gait decline after 5 years. In chapter 5 the association between baseline CSVD markers and the development of parkinsonism after 5 years is studied. Part IV, chapter 6, provides information on the potential clinical and imaging determinants of 8-year mortality in individuals with 3 CSVD. In part V of this thesis, the main results from the studies presented in the preceding chapters are discussed (chapter 7), including possible implications for clinical practice and suggestions for furture research are given in chapter 7. In chapter 8 the main results are summarized and chapter 9 contains a summary in Dutch of this thesis. 4

Part II Cerebral small vessel disease and cognitive performance

2. Cingular integrity and verbal memory performance in cerebral small vessel disease

Published as: H.M. van der Holst*, A.M. Tuladhar*, A.G.W. van Norden, K.F. de Laat, I.W.M. van Uden, L.J.B. van Oudheusden, M.P. Zwiers, D.G. Norris, R.P.C. Kessels, F-E de Leeuw.

Microstructural integrity of the cingulum is related to verbal memory performance in elderly with cerebral small vessel disease: the RUN DMC study. Neuroimage, 2013 Jan;65:416-23

*Both authors contributed equally. CHAPTER 2

Abstract

Background: Cerebral small vessel disease (CSVD) is related to verbal memory failures. It is suggested that early white matter damage, is located, among others, in the (posterior) cingulum at an early stage in neurodegeneration. Changes in the microstructural integrity of the cingulum assessed with diffusion tensor imaging (DTI), beyond detection with conventional magnetic resonance imaging (MRI), may precede macrostructural changes and be related to verbal memory failures. Objective: To investigate the relation between cingular microstructural integrity and verbal memory performance in 503 non-demented older adults with CSVD. Methods: The RUN DMC study is a prospective cohort study in older adults (50-85 years) with CSVD. All participants underwent T1 MPRAGE, FLAIR and DTI scanning and a cognitive test battery, including the Rey Auditory Verbal Learning Test to assess verbal memory performance. Mean diffusivity (MD) and fractional anisotropy (FA) were assessed in six different cingular regions of interests (ROI). Linear regression analysis was used to assess the association between verbal memory performance and cingular DTI parameters, with appropriate adjustments. Furthermore, a Tract-based Spatial Statistics (TBSS) analysis of the whole brain was performed to investigate the specificity of our findings. Results: Both our ROI-based and TBSS analysis showed that FA of the cingulum was positively related to immediate memory, delayed recall, delayed recognition and overall verbal memory performance, independent of confounders. A similar distribution was seen for the inverse association with cingular MD and verbal memory performance with TBSS analysis. No significant relations were found between cingular integrity and psychomotor speed, visuospatial memory and Mini-Mental State Examination (MMSE). When stratified on hippocampal integrity, the MD and FA values of the cingular ROIs differed significantly between participants with good and poor hippocampal integrity; we found lower cingular microstructural integrity in participants with a poor hippocampal integrity. Conclusion: Microstructural integrity of the cingulum, assessed by DTI, is specifically associated with verbal memory performance in older adults with CSVD. Furthermore, we found that when the integrity of the hippocampus is disrupted, the cingular microstructural integrity is impaired as well.

24 CINGULAR INTEGRITY AND VERBAL MEMORY PERFORMANCE IN CEREBRAL SMALL VESSEL DISEASE.

Introduction Cerebral small vessel disease (CSVD) includes, among others, white matter hyperintensities (WMH) and lacunes, and is a frequent finding on magnetic resonance imaging (MRI) scans 1 of elderly people.9 Several patient- and population-based studies, have shown that CSVD is related to verbal memory failure and may eventually result in cognitive decline and dementia in some.26-28 This is thought to be the result of disruption of white matter tracts. White matter damage in Alzheimer dementia has been identified both in post-mortem studies29, 30 as well as 2 in vivo MRI studies.30, 31 A more detailed investigation of the white matter using neuroimaging can be revealed by diffusion tensor imaging (DTI), a non-invasive MRI technique, which provides detailed information on the microstructure and integrity of white matter fibre tracts.32, 33 Two DTI parameters are of special interest: mean diffusivity (MD), a measure of water diffusion 3 averaged in all spatial directions, and fractional anisotropy (FA), which provides information about the directionality of water diffusion. Loss of microstructural integrity is typically accompanied by a decrease in FA and/or an increase in MD.13 There is increasing evidence that DTI parameters are an earlier marker of cognitive decline in comparison to volume measures.34, 4 35 The cingulum bundle, a white matter bundle which connects the medial temporal lobe structures (e.g. hippocampus) and the posterior cingulate cortex, is an important structure for memory function, especially verbal memory performance.36-38 Many DTI studies in mild cognitive impairment (MCI) and Alzheimer dementia have shown that white matter damage, is 5 located, among others, in the (posterior) cingulum at an early stage in neurodegeneration and may be a key marker of early pathology.36, 39-45 To our best knowledge, all studies performed so far had relatively small sample sizes (n<249), used mostly only patients with MCI and Alzheimer dementia, and did not or only limited adjusted for possible confounders. Furthermore, most 6 of these studies used either a region of interest (ROI) approach or a voxel-based morphometry analysis, which have both methodological limitations. Tract-Based Spatial Statistics (TBSS) analysis is a relatively new method, using only those white matter voxels that are in the skeleton (core) of the brains connectional architecture, which enables a robust voxelwise analysis of the 7 microstructural integrity of white matter and this can be accurately matched across subjects.46 In this study, we combined a ROI-based and TBSS approach in order to investigate the association between cingulum integrity and verbal memory performance in non-demented 8 older adults with CSVD. We hypothesized that loss of microstructural integrity of the cingulum is related to impaired verbal memory performance. We examined this in six different ROIs from the posterior to the anterior cingulum in order to get information about regional distribution patterns in respect to verbal memory and performed TBSS analysis of the whole brain as well to 9 investigate the specificity of our findings. In addition, we investigated the role of hippocampal integrity on cingulum integrity in order to assess whether cingular microstructural integrity is associated with hippocampal pathology. This study is part of the Radboud University Nijmegen Diffusion tensor and Magnetic resonance imaging Cohort (RUN DMC) study that included 503 A non-demented, independently living older adults with CSVD, aged between 50 and 85 years.

25 CHAPTER 2

Methods

Study population The RUN DMC study prospectively investigates the risk factors and clinical consequences of brain changes among 503 non-demented older adults with CSVD. The selection procedure of the participants and study protocol were described in detail previously.47 In short, on the basis of established research criteria, CSVD was defined as the presence of lacunes and/or WMH.48 Accordingly, in 2006, consecutive patients referred to the Department of Neurology between October 2002 and November 2006 were selected for participation. Inclusion criteria were: (a) age between 50 and 85 years; (b) CSVD on neuroimaging (WMH and/or lacunes). The main exclusion criteria were dementia,49 (psychiatric) disease interfering with cognitive testing or follow-up, white matter lesions not related to CSVD and MRI contraindications or known claustrophobia. From 1,004 invited individuals by letter, 727 were eligible after contact by phone of whom 525 agreed to participate. In 22 individuals exclusion criteria were found during their visit to our research center, yielding a response of 71.3% (503/705). For the present study, 63 participants were additionally excluded because of territorial infarcts (n=59) and inadequate quality of the MRI images (n=4). All participants signed an informed consent form. The Medical Review Ethics Committee region Arnhem-Nijmegen approved the study.

Measurement of cognitive function Cognitive function was assessed by a standardized neuropsychological test battery performed by two trained investigators (AvN and KdL) and has been described in detail elsewhere.47 For this study, the Mini-Mental State Examination (MMSE) (range 0–30)50 was used as an index of overall cognitive performance. The three-trial version of the Rey Auditory Verbal Learning Test (RAVLT)51 was administered to examine episodic memory formation. To evaluate speed of mental processes we used the Stroop test,52, 53 the Paper-Pencil Memory Scanning Task54 and the Symbol-Digit Substitution Task.55 The Rey Complex Figure Test (RCFT)56 was included as an index of visuospatial memory. We defined four memory indices based on RAVLT performance (overall verbal memory performance, immediate recall, delayed recall, and delayed recognition), as described previously.57, 58 Immediate recall was calculated by the mean of the total number of words remembered in the three learning trials of the RAVLT. Delayed recall was the number of words recalled 30 min after the learning trials. The delayed recognition score was calculated by computing the total of each correctly recognized word (the 15 target words among 15 new distracter items) 30 min after the learning trials. Performance across tests was made comparable by transforming the raw test scores into Z-scores as described elsewhere57 for which the assumption of normality of the distribution was examined. For data reduction purposes, a compound score for overall verbal memory performance was calculated, as

26 CINGULAR INTEGRITY AND VERBAL MEMORY PERFORMANCE IN CEREBRAL SMALL VESSEL DISEASE. described previously,26 by taking the mean of two Z-scores from the RAVLT; one for the added scores on three learning trials of this test and one for the delayed recall of this test. For psychomotor speed and visuospatial memory we also calculated compound scores. 1 Psychomotor speed was calculated as the mean of the Z-scores of the 1-letter subtask of the Paper-Pencil Memory Scanning Task, the reading subtask of the Stroop test and the Symbol-Digit Substitution Task.59 Visuospatial memory is a compound score of the mean of the Z-scores of the immediate recall trial and the delayed recall trial of the RCFT. 2

Conventional MRI Scanning Protocol All participants underwent a cerebral MRI scan on the same 1.5-Tesla Magnetom scanner (Siemens, Erlangen, Germany). The scanning protocol includes whole brain T1-weighted 3 3D magnetization-prepared rapid gradient-echo (MPRAGE) sequence (repetition time (TR)/ echo time (TE)/ inversion time (TI) 2250/3.68/850ms; flip angle 15°; voxelsize 1.0x1.0x1.0mm); a fluid-attenuated inversion recovery (FLAIR) pulse sequences (TR/TE/TI 9000/84/2200ms; voxelsize 1.0x1.2x5.0mm with an interslice gap of 1mm); and a DTI sequence (TR/TE 4 10100/93ms; voxelsize 2.5x2.5x2.5mm; 4 unweighted scans, 30 diffusion weighted scans, with non co-linear orientation of the diffusion-weighting gradient, and b-value 900 s/mm2). For more details of our MRI protocol we refer to our study rationale and protocol published in BioMed Central Neurology in 2011.47 5

Conventional MRI Analysis One experienced investigator, blinded to clinical data (IvU), manually segmented the left and right hippocampus on the MPRAGE image using the interactive software program 6 “ITK-SNAP”.60 Anatomical boundaries were determined in coronal sections with the aid of neuroanatomical atlases,61, 62 and actual segmentation was performed using a previously published protocol63 in which segmentation was performed from posterior to anterior. The details on hippocampus segmentation were described previously in detail.57 Volumes were 7 calculated for the left and right hippocampus separately by summing all voxel volumes of the segmented areas. Inter-rater studies on a random sample of 10% showed an intra-class correlation coefficient for the left hippocampus of 0.73 and for the right hippocampus of 0.79. 8 Intra-rater studies on this sample showed an intra-class correlation coefficient for the left and right hippocampus of 0.97 and 0.96, respectively. We created a mask for the cingulum by applying the white matter atlas (Juelich Histological Atlas white matter labels, provided FMRIB's software library (FSL)) and warped this to the 9 individual T1 volumes (inverse normalization). We counted the number of voxels and multiplied this to voxel volume in order to get the volume of the cingulum. We computed grey and white matter tissue and cerebrospinal fluid (CSF) probability maps using statistical parametric mapping 5 (SPM5) unified segmentation routines on the T1 A MPRAGE images (SPM5; Wellcome Department of Cognitive Neurology, University College London, United Kingdom).64 Total grey and white matter, and CSF volumes were calculated

27 CHAPTER 2 by summing all voxel volumes that had a p>0.5 for belonging to that tissue class. Total brain volume (TBV) was taken as the sum of total grey and white matter. Intracranial volume (ICV) was the summation of all tissue classes (total grey and white matter and CSF volume). To normalize for head size, TBV, hippocampal volume and cingulum volume were expressed as a percentage of total ICV. WMH were manually segmented on FLAIR images and the number of lacunes was rated according to a standardized protocol.47 In a random sample of 10%, inter-rater variability for total WMH volume yielded an intra-class correlation coefficient of 0.99, intra- and inter-rater reliability for the lacunes yielded a weighted kappa of 0.80 and 0.88. D C B E A DTI Analysis Diffusion data were first pre-processed to detect and correct head and cardiac motion artefacts, using an in-house developed iteratively re-weighted-least-squares algorithm named ‘PATCH’ (www.ru.nl/neuroimaging/diffusion).65 Corrections of Eddy current and motion artefacts from affine misalignment were performed simultaneously by minimization of the residual diffusion tensor errors.66 Next, FA and MD images were calculated using a DTIFit within the Functional MR of the Brain diffusion toolbox, which were then fed into the TBSS pipeline.46 The thinning procedure was conducted on the mean FA image to create a common skeleton, which represents the core-structure of the white matter tract. Subsequently this skeleton was thresholded at FA-value 0.3 to include the major white matter tracts and to account for the inter-subject variability. All normalized FA data were then projected onto this skeleton. These skeleton projection factors were then applied to the mean images. These data were then fed into the voxel-wise cross-subject statistics. In addition, MD and FA values of the cingulum were measured in six bilateral ROIs (A-F) in the cingulum bundle as illustrated in figure 1. An experienced neurologist (FEdL) blinded to subject information used FSLView67 to overlay an histological atlas (Juelich Histological Atlas)68 over the MNI152 standard brain, in order to manually mark out the centre coordinates of the ROIs in the different locations of the cingulum and the parahippocampal region (based on neuroanatomy data from the literature).69 We marked out 5 ROIs on a sagittal slice of the cingulum. The cingulum was divided in 4 parts: anterior, middle, posterior and parahippocampal section.39 The centre of ROI C was placed at the centre of the middle curve of the cingulum fibres, just above the body of the corpus callosum: this is the middle cingulate region. We placed the centre of the 2 anterior ROIs (ROI A and B) before the middle curve of the cingulum (around the rostrum and genu of the corpus callosum) and the 2 posterior ROIs (ROI D and E) nearby the dorsal curve of the cingulum fibres (around the splenium of the corpus callosum). The centre of the sixth ROI (ROI F) was placed in the parahippocampal cingulum on the medial temporal portion of the cingular fibres. The centre coordinates of each ROI were mirrored with respect to the mid-sagittal plane to obtain these same regions in both cerebral hemispheres. The ROI centre coordinates were mapped back to the individual DTI space of each subject using the inverse of the SPM5 unified T1 normalization parameters.64 ROIs were defined as 6 mm diameter spheres

28 CINGULAR INTEGRITY AND VERBAL MEMORY PERFORMANCE IN CEREBRAL SMALL VESSEL DISEASE.

(volume of 0.056 mL) around the DTI-space centre coordinates. We used a modest ROI size to ensure that all ROIs included only white matter (which was visually checked). FA and MD values within each ROI were averaged. Furthermore, the mean MD was calculated in both hippocampi. 1 All images were visually checked for motion artefacts and co-registration errors, especially for not including perihippocampal CSF.

D C B E A 3

Figure 1. Location of the six (A-F) regions of interest in the left cingulum in a sagittal plane. 5 Other measurements The following characteristics were considered possible confounders: age, sex, educational level70 and depressive symptoms. Depressive symptoms were assessed using the Centre of Epidemiologic Studies on Depression Scale (CES-D).71 6

Structural integrity of the hippocampus In order to investigate the role of hippocampal integrity on the cingular microstructural integrity, we composed a score for structural hippocampal integrity. As it has been 7 demonstrated that a combination of high diffusivity of the hippocampus and low hippocampal volume is related to conversion to Alzheimer dementia in patients with MCI,34 we compiled a score for hippocampal integrity using both parameters. Both diffusivity and 8 volume were given a score, ranging from 1 (poor: lowest tertile of the volume/highest tertile of MD distribution) to 3 (good; highest tertile of the volume/lowest tertile of the MD distribution) leading to a maximum score between 2 (lowest tertile of the volume and highest tertile of the MD distribution (1+1): worst hippocampal integrity) and 6 (highest tertile of the volume and 9 lowest tertile of the MD distribution (3+3):best hippocampal integrity). A good hippocampal integrity was defined as a score of ≥ 4, while a poor integrity was defined as a score of 2 or 3.

29 CHAPTER 2

Statistical analysis Statistical analyses were performed with SPSS 16.0 for Windows (SPSS Inc., Chicago, IL, USA). Baseline characteristics were summarized as mean (standard deviation (SD)) or proportions; for skewed variability parameters the median and the interquartile range were calculated. For the TBSS analysis, we assessed voxel-wise correlations between the skeletal DTI parameters (FA and MD) and cognitive performance, including the four different indices of the RAVLT performance, psychomotor speed, visuospatial memory and MMSE. Adjustments were made for potential confounders including age, sex, educational level, depressive symptoms and normalized total brain. We applied a permutation-based statistical interference tool for non-parametric approach as a part of the Functional MRI of the Brain Software Library, with number of permutation tests set to 5000.46, 72 Significant clusters were identified using the threshold-free cluster enhancement with a p<0.05, corrected for multiple comparisons.73 In addition, we performed the same analysis with a threshold at p<0.01. For the ROI analyses, we computed regression coefficients of the mean FA and MD of the six ROIs in the cingulum with the same cognitive test as mentioned above. Adjustments were made for the same confounders as mentioned above. Regression coefficients are presented as standardized β-values. To assess whether the structural integrity of the hippocampus modified the relation between cingulum microstructural integrity and performance in verbal memory, psychomotor speed, visuospatial memory and MMSE, we performed the previously described linear regression analyses stratified in two groups of hippocampal integrity (good and poor). Adjustments were made for the same confounders as mentioned above. To assess whether the groups differed significantly, the correlation coefficients were converted to Z-scores using the Fisher r-to-z transformation. The obtained z-scores were used to calculate the p-values. In addition, we performed independent two-sample t-test to assess the differences in MD and FA values of the cingulum when stratified on hippocampal integrity in order to assess whether cingular microstructural integrity is associated with hippocampal pathology. Bonferroni corrections were applied to correct for multiple comparisons (p<0.00385 were considered significant).

30 CINGULAR INTEGRITY AND VERBAL MEMORY PERFORMANCE IN CEREBRAL SMALL VESSEL DISEASE.

Results The baseline characteristics of the 440 participants are shown in Table 1. The mean age of the population was 65.2 years (SD 8.9) and 54.3% were male. The mean MMSE score was 28.2 (SD 1 1.6). In Table 2 the test scores of the cognitive test battery are shown.

Table 1. Baseline characteristics of the 440 participants 2 Baseline characteristics n = 440 Age, mean (SD), years 65.2 (8.9) Male sex, No. (%) 239 (54.3) Participants with only primary education, No. (%) 41 (9.3) 3 MMSE score, mean (SD) 28.2 (1.6) CES-D score, mean (SD)a 11.1 (9.4) Participants with depressive symptomsb, No. (%) 151 (34.3) Imaging characteristics 4 Total brain volume, mean (SD), mL 1098.6 (119.0) Intracranial volume, mean (SD), mL 1674.7 (157.4) Normalized hippocampal volumec, mean (SD) 0.41 (0.06) Normalized cingular volumec, mean (SD) 0.29 (0.02) 5 White matter volume, mean (SD), mL 466.8 (66.9) WMH volume, median (IQR), mL 6.5 (3.2-17.9) Lacunes presence, No. (%) 135 (30.7) 6 Abbreviations: CES-D: Centre of Epidemiological Studies on Depression Scale; IQR: interquartile range; MMSE: Minimal Mental State Examination; SD: standard deviation; WMH: White Matter Hyperintensity. a Two participants had missing values on CES-D score. b Defined as CES-D scores ≥16 and/or the current use of anti-depressive medication. 7 c Volumes are represented as percentage of total intracranial volume (SD).

31 CHAPTER 2

Table 2. Cognitive test scores

RAVLT: number of words recalled n = 439 Immediate recall trial 1 5.2 (1.8) Immediate recall trial 2 7.4 (2.3) Immediate recall trial 3 8.7 (2.6) Total of immediate recall trial 1-3 21.3 (6.1) Delayed recall 6.0 (3.1) Delayed recognition 27.1 (3.2) a Stroop test (time in seconds) n = 427 Trial 1 (words) 25.5 (6.1) Trial 2 (colors) 32.9 (7.5) Trial 3 (concept-shifting) 62.8 (20.9) Paper and pencil memory scanning task (time in seconds) n = 428 1 character 44.3 (13.5) 2 characters 61.2 (18.7) 3 characters 76.0 (26.2) Symbol digit substitution task (No. in 60s) n = 437 Total score 27.7 (9.7) Rey complex figure test (range 0-36) n = 435 Copy trial 33.5 (3.4) Immediate recall trial 18.1 (6.8) b Delayed recall trial 18.1 (6.7) c

Abbreviations: RAVLT: Rey Auditory Verbal Learning Test. Numbers represent mean (standard deviation). a 3 participants had missing values on delayed recognition. b 5 participants had missing values on immediate recall trial. c 19 participants had missing values on delayed recall trial.

32 CINGULAR INTEGRITY AND VERBAL MEMORY PERFORMANCE IN CEREBRAL SMALL VESSEL DISEASE.

Figure 2 shows the association between the voxel-wise analysis of the FA and verbal memory performance (p<0.05, corrected for multiple comparisons). For this analysis (and the ROI analysis) we had to exclude 2 participants because of missing data on the CES-D. FA in the 1 cingulum and corpus callosum were positively associated with immediate memory, delayed recall, delayed recognition and overall verbal memory performance, especially in the left hemisphere and independent of confounders. By contrast, we found no significant relations with psychomotor speed and neither with visuospatial memory and MMSE (data not shown). 2 We found a similar distribution for the inverse association with MD and verbal memory performance, although these associations were less significant (data are not shown).

Figure 2. Voxel-wise analyses of the fractional anisotropy values and verbal memory performance and psychomotor speed. 8 Voxel-wise analyses of fractional anisotropy (FA) values were positively associated with the four memory indices of the Rey Auditory Verbal Learning Test (RAVLT), including immediate memory, delayed recall, delayed recognition and overall verbal memory performance. No significant associations were found for FA values and psychomotor speed. Adjustments were made for age, sex, educational level, depressive symptoms, normalized total brain volume, normalized hippocampal volume, white matter hyperintensity volume and the number of lacunes. 9

Our ROI analyses show similar results. In Table 3 the associations between the FA of the six ROIs in left and right cingulum and the different cognitive tests are shown. Significant A associations were found between the FA values in the left mid cingulum (ROI C) and immediate memory (β=0.09, p=0.041) and delayed recognition (β=0.11, p=0.027); in the

33 CHAPTER 2 left ROI D and overall verbal memory performance (β=0.10, p=0.027), immediate memory (β=0.14, p=0.003) and delayed recognition (β=0.10, p=0.043); in the right ROI D and overall verbal memory performance (β=0.10, p=0.025) and immediate memory (β=0.10, p=0.018); in the right ROI E and overall verbal memory performance (β=0.09, p=0.037) and immediate memory (β=0.11, p=0.009), independent of confounders. No significant associations were found with psychomotor speed, visuospatial memory and MMSE. Almost no significant results were found between MD values and verbal memory performance (data not shown). When we stratified the above mentioned relation on a good (n=291) and poor (n=147) hippocampal integrity we did not find significant differences in the correlations coefficients between both groups, except for the relation with immediate memory and FA in the left ROI B (results are not shown). Table 4 shows cingulum integrity (expressed as mean MD and FA values of the six ROIs in the left and right cingulum) stratified on hippocampal integrity. The MD and FA values of all ROIs differed significantly between both groups (good and poor hippocampal integrity), except for the FA values in ROI E and F. Our results showed that the cingular microstructural integrity is significantly lower in participants with a poor hippocampal integrity than in participants with a good hippocampal integrity.

34 CINGULAR INTEGRITY AND VERBAL MEMORY PERFORMANCE IN CEREBRAL SMALL VESSEL DISEASE.

Right -.02 -.04 .01 -.04 -.07 -.05 .01 1 FA ROI F FA Left -.01 -.01 -.01 .05 -.02 .01 -.05

c b 2 Right .09 .11 .06 .03 .05 .04 .03 FA ROI E FA Left .06 .07 .05 .06 -.04 .02 .02 3 c c Right .10 .10 .08 .09 -.02 .06 .01 c b,d c To posterior To

à ROI D FA Left .10 .14 .06 .10 .01 .08 .01 4 Right -.01 -.01 .01 .07 -.08 -.03 .02

c c 5 Mid cingulum ROI C FA Left .08 .09 .05 .11 .01 .03 .03

region of interest region ß ROI: Right -.01 -.02 .01 .04 -.06 -.07 -.05 6 To anterior anterior To ROI B FA Left .04 .05 .03 .06 -.06 .07 -.01

Right -.01 .01 -.01 .05 -.09 -.04 -.06 7 FA ROI A FA Left .03 .06 .01 .06 -.09 -.01 -.06 mini-mental state examination; examination; state mini-mental a 8 MMSE: a

a 9 fractional anisotropy; anisotropy; fractional FA: Immediate memory Immediate recall Delayed recognition Delayed A <0.05. <0.01. p Overall verbal memory performance verbal Overall speed Psychomotor memory Visuospatial MMSE Significant after Bonferroni correction. Bonferroni after Significant p Compound scores. c Table 3. Association between the fractional anisotropy values of the six bilateral regions of interest in the cingulum and cognitive performance. the cingulum and cognitive in interest of regions the six bilateral of values anisotropy fractional the between Association 3. Table Abbreviations: volume, brain total normalized symptoms, level, depressive sex, educational for age, adjusted betas), (standardized coefficients regression represent Numbers and the number of lacunes. volume hyperintensity matter white volume, hippocampal normalized a b d

35 CHAPTER 2

Table 4. Mean MD and FA of the six bilateral regions of interest in the cingulum stratified on hippocampal integrity

Hippocampal integrity

Good Poor Good Poor MD MD p-valuea FA FA p-valuea ROI A: anterior Left 8.29 (0.74) 8.83 (1.90) <0.01 0.41 (0.10) 0.38 (0.10) <0.05 Right 8.22 (0.60) 8.75 (1.91) <0.01 0.39 (0.08) 0.37 (0.08) <0.01 ROI B: anterior Left 8.24 (0.93) 8.82 (1.69) <0.001 0.42 (0.10) 0.37 (0.10) <0.001 Right 8.12 (0.67) 8.85 (1.82) <0.001 0.42 (0.09) 0.39 (0.09) <0.001 ROI C: mid cingulum Left 7.68 (0.55) 8.32 (1.76) <0.001 0.53 (0.09) 0.47 (0.10) <0.001 Right 7.78 (0.55) 8.40 (1.54) <0.001 0.52 (0.09) 0.47 (0.09) <0.001 ROI D: superior-posterior Left 7.80 (0.42) 8.28 (1.84) <0.01 0.54 (0.07) 0.51 (0.09) <0.001 Right 7.76 (0.48) 8.21 (1.63) <0.01 0.51 (0.07) 0.49 (0.07) <0.01 ROI E: inferior-posterior Left 7.82 (0.47) 8.10 (1.80) <0.05 0.45 (0.07) 0.46 (0.07) NS Right 7.98 (0.55) 8.41 1.71) <0.001 0.39 (0.07) 0.38 (0.07) NS ROI F: parahippocampal Left 8.57 (0.74) 9.01 (0.81) <0.001 0.25 (0.06) 0.25 (0.07) NS Right 8.74 (0.71) 9.27 (1.00) <0.001 0.25 (0.08) 0.25 (0.07) NS

Abbreviations: FA: Fractional Anisotropy; MD: Mean Diffusivity, expressed as x10-4 mm2/s; NS: Not Significant; ROI: region of interest Numbers represent mean (standard deviation) a p-value for difference between both groups (good and poor hippocampal quality) calculaed with a two sample t-test

36 CINGULAR INTEGRITY AND VERBAL MEMORY PERFORMANCE IN CEREBRAL SMALL VESSEL DISEASE.

Discussion In this study we investigated the association between cingulum microstructural integrity and verbal memory performance by as well an ROI based approach and a TBSS analysis 1 of the cingulum in 503 non-demented older adults with CSVD. We demonstrated that the microstructural integrity of the cingulum is specifically associated with verbal memory performance, independent of confounders, including hippocampal volume and coexisting CSVD. Furthermore, we found that when the integrity of the hippocampus is disrupted, the 2 microstructural integrity of the cingulum is impaired as well. To the best of our knowledge, this is the first study investigating the association between the microstructural integrity of the cingulum and verbal memory function in a large group of non- demented, independently living elderly with CSVD. Our data show that the microstructural 3 integrity of the cingulum beyond the detection limit of conventional MRI is associated with episodic memory formation. On conventional MRI only less than 6% of the ROIs in the cingulum of our study population showed the presence of visible WMH, strengthening our assumption that we indeed investigated the earliest structural changes in the cingulum 4 bundle integrity. Our finding is in line with the results of previous studies in patients with MCI or Alzheimer dementia showing that loss of structural integrity of the cingulum is associated with impaired verbal memory performance.36, 38, 74-76 5 Furthermore, we found that participants with a poor hippocampal integrity (low volume and high MD) had a significantly lower microstructural integrity of the cingulum (high MD as well as low FA) than participants with a good hippocampal integrity (high volume and low MD). In addition, we found that the association between cingulum integrity and verbal memory 6 performance did not differ significantly between both groups (good and poor hippocampal integrity). These results might suggest that hippocampal integrity is not an intermediate in this association. There is still much debate whether, in general, white matter pathology is related to, or independent of, grey matter pathology. There are three main hypotheses, 7 one suggesting that microstructural white matter changes occur as a result of Wallerian degeneration,77 meaning that white matter pathology is preceded by grey matter pathology. In contrast, the retrogenesis theory proposes that loss of white matter integrity is the result of 8 myelin breakdown that occurs in the reverse order to myelogenesis.78, 79 Furthermore, there is evidence that CSVD can cause white matter damage occurring as a result of oligodendrocyte death and reactive gliosis.80 Our results are not able to distinguish between these theories, also because of the cross-sectional design of our study. 9 With our TBSS analysis we found that the microstructural integrity of the mid-anterior part of the cingulum is associated with impaired verbal memory, while our less precise ROI analysis found the mid-posterior part to be involved. As mentioned before, most studies in MCI and Alzheimer dementia found mainly involvement of the posterior cingulum. However, A most of them included only a few ROIs in the cingulum, did not perform a TBSS analysis on the cingulum and investigate the cingulum integrity in patients with already MCI and/or

37 CHAPTER 2

Alzheimer dementia. Probably, our results suggest that loss of structural integrity starts in the middle-anterior parts of the cingulum and maybe further spreading to the posterior parts, near medial temporal lobe structures, with progression of verbal memory impairment and ultimately dementia. Future studies are needed to describe the possible spread over time of the pathological process throughout the cingulum. A follow-up of the RUN DMC study is already being executed to further investigate this. Our findings show that cingulum integrity is specifically associated with verbal episodic memory. We did not find relations between cingulum integrity and psychomotor speed, visuospatial memory and MMSE. Although a recent study found associations between (posterior) cingulum integrity and attention/executive functioning and visuospatial memory,75 this discrepancy might be due to the limited additional adjustments performed in the latter study. Furthermore, their study population differed from ours. Of their 220 participants, 149 had MCI, while our study population had a mean MMSE of 28.1 and only a small percentage (10%) sufficed the diagnosis MCI according to the definition of van der Elst.51 As a result, it is likely that more damage to the structural integrity of the cingulum and medial temporal lobe structures like the hippocampus has occurred in the study population of Kantarci.75 The reason for not finding a relation between DTI parameters and visuospatial memory in our study may be due to the fact that memory performance based on the RCFT may in part rely on motor learning. Therefore, the RCFT cannot be regarded as a pure measure of spatial memory, as part of episodic memory, which as a result may be less dependent of the hippocampal memory circuit. In our study also the corpus callosum is involved in verbal memory performance, as shown by our TBSS analysis. DTI abnormalities in MCI and Alzheimer dementia have been found in the corpus callosum.81, 82 A study of Christman and Propper83 showed that episodic memory is at least in part dependent on interhemispheric interaction. Furthermore, it has been reported that older adults show reduced hemispheric asymmetry in episodic memory functioning.84 Disruption of the corpus callosum can lead to impaired interhemispheric interaction, resulting in impaired (verbal) memory performance. With respect to lateralization effects, our study showed that the strongest association with verbal memory in the left hemisphere. As would be expected from a verbal task, the dominant (left) hemisphere, containing the language centres (Broca and Wernicke), is more strongly involved in performing the RALVT, as most of our participants were right-handed. Still, a few methodological considerations need to be addressed. Although our data are derived from the largest DTI study on the cingulum thus far, our study design is cross-sectional, which prevents us from drawing conclusions with respect to causality. Data of our follow-up study,47 which is currently underway, might help to further investigate the structural integrity of the cingulum in relation to structural changes in the medial temporal lobe structures and in the development of cognitive decline and dementia. Furthermore, we used an atlas to identify the cingulum bundle, and manually placed the ROIs in order to avoid grey matter and white matter of other tracts to blur our DTI measures. Although the placement of the

38 CINGULAR INTEGRITY AND VERBAL MEMORY PERFORMANCE IN CEREBRAL SMALL VESSEL DISEASE. atlas was visually checked and the placement of the ROIs was performed by an experienced neurologist, blinded to clinical data, this method can be somewhat subjective and can make interpretation between studies difficult. 1 A major strength of our study is the fact that it is a large, single-centre study with a response rate of over 70%. All MRI data were acquired on a single scanner in a similar way, and the hippocampus and WMH were assessed volumetrically in a reliable, sensitive and objective way by two trained experts, who were blinded to all clinical data. Furthermore, extensive 2 adjustments were made for possible confounders. Moreover, cognitive function was assessed by only two investigators using a standardized cognitive battery. In conclusion, this study demonstrates that microstructural integrity of the cingulum, assessed by TBSS analysis and a ROI-based approach, is specifically associated with episodic memory 3 function, notably verbal memory, in non-demented older adults with CSVD. Furthermore, we showed that the microstructural integrity of the cingulum is significantly lower in participants with a poor hippocampal integrity than in those with a good hippocampal integrity. Consequently, our findings show that DTI, by using a ROI-based and TBSS approach, is a 4 sensitive diagnostic tool to detect early microstructural changes in the cingulum, which are related to impaired verbal memory, before structural changes of cingulum can be detected on conventional MRI. Future studies should prospectively investigate the predictive value of DTI parameters of the cingulum in relation to cognitive consequences of CSVD and incident 5 dementia. If the predictive value is proven, DTI of the cingulum could possibly be a surrogate marker for development of cognitive decline and dementia and could be a starting point for therapeutic trials aiming to prevent disease progression. 6

Part III Cerebral small vessel disease and motor performance

3. Baseline cerebral small vessel disease and gait decline

Published as: H.M. van der Holst, I.W.M. van Uden, K.F. de Laat, E.M.C. van Leijsen, A.G.W. van Norden, D.G. Norris, E.J. van Dijk, A.M. Tuladhar, F-E de Leeuw.

Baseline cerebral small vessel disease is not associated with gait decline after 5 years. Movement Disorders Clinical Practice, 2016 Nov. DOI:10.1002/mdc3.12435 CHAPTER 3

Abstract

Background: Cerebral small vessel disease (CSVD) is cross-sectionally associated with gait disturbances; however, the relation between baseline CSVD and gait decline over time is uncertain. Furthermore, diffusion tensor imaging (DTI) studies on gait decline are currently lacking. Objective: To investigate the association between baseline imaging CSVD markers and gait decline. Methods: In 2006, 310 participants from the RUN DMC cohort, a prospective cohort with older adults aged 50-85 years with CSVD, were included. Gait variables were assessed using a computerized walkway (GAITRite) during baseline and follow-up. Linear and logistic regression analyses were used to investigate the relation between imaging measures and gait decline and incident gait impairment (speed ≤ 1.0m/s). Tract-based spatial statistics (TBSS) analysis was used to identify possible differences in DTI measures of white matter tracts between participants with and without incident gait impairment. Results: Mean age was 63.3 years (SD 8.4) and mean follow-up duration 5.4 years (SD 0.2). No significant associations between imaging measures and gait decline were found. TBSS analysis revealed no significant differences in DTI measures between participants with and without incident gait impairment after additional adjustment for SVD. In sub- analyses, a high total white matter hyperintensity (WMH) volume (OR 2.8 for highest quartile, 95% CI: 1.1-7.1) and high infratentorial WMH volume (OR 1.8 per SD increase, 95% CI: 1.1-2.9) were associated with an increased 5-year risk of gait impairment, although this was not significant after correction for multiple testing. Conclusion: Baseline imaging CSVD markers were not associated with gait decline or incident gait impairment after 5 years. Future studies should investigate whether CSVD progression is related to gait deterioration.

44 BASELINE CEREBRAL SMALL VESSEL DISEASE AND GAIT DECLINE.

Introduction Gait impairment has a major impact on the quality of life of older adults and is associated with adverse outcomes including decline in activities of daily living, falls, cognitive impairment, 1 hospitalization and death.25, 85, 86 Cerebral small vessel disease (CSVD) has been identified as a possible risk factor of gait impairment, albeit mainly in cross-sectional studies.16, 17 Few studies have investigated the relation between CSVD and gait decline over time, often only by taking white matter hyperintensities (WMH) into account, whereas the spectrum of traditional 2 CSVD markers also includes lacunes, microbleeds and brain atrophy. These previous studies showed conflicting results, reporting no,87 or weak positive associations,88-90 of which some found a dose-dependent effect,89, 90 while others postulated a threshold effect of WMH88 after which gait decline became apparent. 3 Possibly, the underlying microstructural integrity of the white matter (WM), which can be assessed by diffusion tensor imaging (DTI), plays a role in gait decline. It has been suggested that changes in WM integrity precede the development of WMH.91 Gait impairment has been cross-sectionally associated with WM integrity,92 however, studies on gait decline using DTI 4 are currently lacking. There is limited evidence about a clinical relevant change in gait speed.85 However, a gait speed <1.0m/s has been consistently associated with major adverse health-related outcomes in well-functioning older adults,86 which might therefore be a clinical useful cut point for the 5 development of gait impairment in clinical practice. To our knowledge, no previous studies have investigated the relation between CSVD and the development of gait impairment (speed <1.0m/s). The aim of this study was to investigate whether baseline CSVD, including conventional MRI 6 and DTI markers of CSVD, is associated with gait decline and incident gait impairment after 5 years of follow-up. This study may provide insight into the role of CSVD and gait deterioration and could possibly help to identify adults with CSVD at highest risk for gait decline and incident gait impairment. 7

45 CHAPTER 3

Methods Study population This study is part of the Radboud University Nijmegen Diffusion tensor and Magnetic resonance Cohort study (RUN DMC study), which studies the risk factors and clinical consequences of brain changes as assessed by MRI in 503 participants with CSVD. The recruitment, study rationale and protocol of the RUN DMC study have been described in detail elsewhere.47 A CSVD diagnosis was made based on the results of brain imaging and included the presence of WMH and/or lacunes of presumed vascular origin.48 In 2006, baseline data collection was performed. Inclusion criteria were age 50-85 years and CSVD on brain imaging. Main exclusion criteria were: parkinsonism, dementia, life expectancy <6 months, non-CSVD related WM lesions (e.g. multiple sclerosis), and MRI contra-indications.47 Follow-up assessment was performed in 2011-2012. Of the 503 baseline participants, 398 participated in the follow-up examination. For the present study, we excluded 88 participants, yielding a final sample of 310 (see flowchart Figure 1). All participants signed an informed consent form. The Medical Review Ethics Committee region Arnhem-Nijmegen approved the study.

Gait measurement and gait impairment Quantitative gait analysis was performed by using a 5.6 meter electronic portable walkway (GAITRite, MAP/CIR Inc., Havertown, PA), which has an excellent test-retest reliability and validity.93, 94 Each participant was instructed to walk twice over the walkway at a self selected usual gait speed. In order to measure steady-state walking, participants started two meters before the walkway and stopped two meters behind it. The following gait parameters were averaged over two walks: gait speed (m/s) and its components stride length (m) (the distance between the heel points of two consecutive footprints of the same foot) and cadence (number of steps per minute). Changes over time in these gait parameters were calculated as the difference between follow-up and baseline assessment. We considered a gait speed decline of ≥ 0.1m/s as a significant decline.25, 95 Gait impairment was defined as a gait speed <1.0m/s.86

46 BASELINE CEREBRAL SMALL VESSEL DISEASE AND GAIT DECLINE.

Baseline study population n = 503 1

Lost to follow-up n = 2

Deceased n = 49 2

Unable to visit research centre n = 54 • Illness that prevented visit (19) • Moved (5) • Lack of time (30) Follow-up study population 3 n = 398

Exclusion criteria n = 88

• Imaging artefacts at baseline (1) 4 •Territorial infarcts at baseline (40) •Parkinsonism at follow-up (12) •Conditions associated with gait impairment other than CSVD at baseline and follow-up (25) •Missing gait data at follow-up (10) 5 Participants for analysis n = 310

Figure 1. Flowchart of the study population 6 Abbreviations: CSVD: cerebral small vessel disease Of the 503 baseline participants, 2 participants were lost to follow-up, 49 had died and 54 refused an in-person follow-up examination, but their clinical endpoints were available; 398 participated in the follow-up assessment. For the present study, we included 310 participants, 88 participants were additionally excluded because of (i) 7 baseline T1-T2 artefacts (n=1), (ii) territorial infarcts at baseline imaging (n=40), because these infarcts were considered as a potential confounder, (iii) parkinsonism during follow-up examination (n=12), because apart from CSVD other pathologies as amyloid pathology, Lewy body pathology and nigrastriatal dopaminergic loss could play a role in gait deterioration in these patients, (iv) other conditions than CSVD associated with gait 8 impairment which prevent participants from walking unaided at baseline and follow-up (n=25) (e.g. joint fusion, severe arthritis, severe polyneuropathy, leg amputation, severe vision problems, severe cardiac or respiratory diseases, severe peripheral arterial disease and psychogenic gait disturbance) and (v) missing data on follow-up GAITRite (n=10) (because they were wheelchair bound, because of home visit or technical problems), yielding a final sample of 310 participants. 9

47 CHAPTER 3

MRI protocol All participants underwent a cerebral MRI on a 1.5-Tesla Magnetom Sonata scanner (Siemens Medical Solutions, Erlangen, Germany) at baseline. The protocol included the following scans: a T1-weighted 3D magnetization-prepared rapid gradient-echo (MPRAGE) imaging (repetition time (TR)=2250ms, echo time (TE)=3.68ms, inversion time (TI)=850ms, flip angle=15°, voxelsize 1.0x1.0x1.0mm); a Fluid-attenuated inversion recovery (FLAIR) sequence (TR=9000ms, TE=84ms, TI=2200ms, voxelsize 1.0x1.2x5.0mm, with an interslice gap of 1mm); a transversal T2*weighted gradient echo sequence (TR=800ms, TE=26ms, voxelsize 1.3x1.0x5.0mm with an interslice gap of 1mm) and a DTI sequence (TR=10100ms, TE=93ms, voxelsize 2.5x2.5x2.5mm; 4 unweighted scans, 30 diffusion weighted scans with b-value=900 s/mm²).

MRI analysis WMH were manually segmented on the FLAIR images and total WMH volume was calculated by summing all segmented areas multiplied by slice thickness, with a good inter-rater variability (intraclass correlation coefficient: 0.99). WMH were also determined in predefined regions taken from an inversely normalized Talairach-based atlas,96 and included frontal, parietal, occipital, temporal lobes and sublobar (basal ganglia, thalamus, internal and external capsule, insula), limbic (cingulate gyrus) and infratentorial (brainstem and cerebellum) areas. The ratings of lacunes and microbleeds were rated according to the recently published Standards for Reporting Vascular changes on neuroimaging (STRIVE)5 by trained raters blinded to clinical information (intra-rater and inter-rater reliabilities: for lacunes: weighted kappa values 0.87 and 0.95, respectively, and for microbleeds: 0.85 and 0.86, respectively).97 To obtain grey matter (GM) and WM and cerebrospinal fluid (CSF) volume automated segmentation on T1 images was done using Statistical Parametric Mapping 12 unified segmentation routines (SPM12; Wellcome Department of Cognitive Neurology, University College London, United Kingdom; http://www.fil.ion.ucl.ac.uk/spm/software/spm12/). The volumes were calculated by summing all the voxel volumes belonging to that tissue class. All images were visually checked for co-registration errors and motion and/or segmentation artefacts. All volumes were normalized to the total intracranial volume (sum of GM, WM and CSF)98 to adjust for head size. GM volume was composed of the volume of the neocortex, basal ganglia and thalamus.

DTI analysis Diffusion data were pre-processed and analyzed according to a previous described procedure.47 The diffusion-weighted images of each participant were realigned on the mean of the unweighted image using mutual information based co-registration routines from SPM5. The diffusion tensor32 and its eigenvalues were estimated using linear regression using an SPM5 add-on (http://sourceforge.net/projects/spmtools). Spurious negative eigenvalues were set to zero, after which the tensor derivates fractional anisotropy (FA) and mean

48 BASELINE CEREBRAL SMALL VESSEL DISEASE AND GAIT DECLINE. diffusivity (MD) were calculated.99 The mean unweighted image was used to compute the co- registration parameters to the anatomic T1 reference image, which were then applied to all diffusion-weighted images and results. All images were visually checked for motion artefacts 1 and co-registration errors. The mean FA and MD were then calculated in the total WM. For the tract-based spatial statistics (TBSS) analysis, DTIFit within the FSL toolbox was used to generate FA and MD images, which were imported into the TBSS pipeline.46 To create a FA skeleton, the mean FA image was thinned and subsequently this skeleton was thresholded 2 at 0.3 to include major WM tracts.

Cognition and other measurements Global cognitive function was evaluated by the Mini Mental State Examination (MMSE)50 and 3 the Cognitive Index, a constructed compound score. The cognitive index was calculated as the mean of the z-scores of the Speed Accuracy Tradeoff (SAT) score of the 1-letter subtask of the Paper-Pencil Memory Scanning Task, the mean of the SAT score of the reading task of the Stroop test, the mean of the Symbol-Digit Substitution task and the mean of the added 4 score on the three learning trials of the Rey Auditory Verbal learning test and the mean of the delayed recall of this test.100 To adjust for the number of faults in the Paper-Pencil Memory Scanning Task and the Stroop test, we used SAT scores (accuracy (%)/reaction time). Barthel index (range 0-20) was used to assess functional independence.101 5

Statistical analysis Statistical analyses were performed with IBM SPSS Statistics 20 for Windows. To compare the baseline characteristics between participants who were included in this 6 study and those who dropped out, age and sex-adjusted ANCOVA or logistic regression were used. Gait parameters assessed during baseline and follow-up were compared by using paired t-tests. The associations between baseline imaging measures and changes in gait parameters were assessed using multiple linear regression analysis. Adjustments were made 7 for follow-up duration and baseline age, sex, height, gait parameters, cognitive index and GM volume (when investigating traditional CSVD markers and DTI measures) and/or traditional CSVD markers (when investigating GM volume and DTI measures). WMH volume was log 8 transformed, because of the skewed distribution. To ensure that multicollinearity was not present, variance inflation factor (VIF) was calculated for all regression models presented. The VIF scores were low (<3) for all models (VIF-scores >5 are considered to reflect high multicollinearity). Data were presented as standardized betas. 9 Logistic regression analysis was used to calculate odds ratios (OR) and 95% confidence intervals (CI) to quantify the relation between baseline imaging measures and incident gait impairment (<1.0m/s), adjusted for the same confounders as described above. p Results with a <0.05 were considered significant. Bonferroni corrections were used to A correct for multiple testing.

49 CHAPTER 3

To compare voxel-wise analyses of DTI measures (FA and MD) between participants with incident gait impairment (n=48) and those without gait impairment (n=240) a two-sample t-test was performed, using a permutation-based statistical interference as part of FSL toolbox (‘randomize’), with a standard number of permutation tests set a 5000. Adjustments were made for follow-up duration, baseline age, sex, height, gait speed, cognitive index and total brain volume and additionally for traditional CSVD markers. Four participants were excluded because of missing values of microbleeds and DTI artefacts.

Table 1. Baseline characteristics of the study sample

Participants Participants not p-value for Characteristics included included difference Demographics n = 310 n = 193 Age, mean (SD), years 63.3 (8.4) 69.5 (8.1) <0.001a Male sex, No. (%) 173 (55.8) 111 (57.5) 0.80b MMSE score, mean (SD) 28.4 (1.5) 27.7 (1.8) 0.005a Cognitive index, mean (SD) 0.24 (0.74) -0.42 (0.67)c <0.001a Barthel index, mean (SD) 19.8 (0.5) 19.5 (1.2) 0.002a Gait characteristics n = 310 n = 189d Gait speed, mean (SD), m/s 1.37 (0.22) 1.13 (0.29) <0.001a Gait impairment (gait speed <1.0m/s), No. (%) 18 (5.8) 52 (27.5) <0.001b Imaging measures e n = 310 n = 192f WMH volume, median (IQR), mL 5.1 (2.9-12.0) 13.3 (5.9-25.5) <0.001a Lacunes, presence, No. (%) 56 (18.1) 78 (40.6) <0.001b Microbleeds, presenceg, No. (%) 43 (14.0) 38 (19.9) 0.82b WM volume, mean (SD), mL 472.9 (37.9) 450.6 (50.3) 0.02a GM volume, mean (SD), mL 628.4 (46.9) 596.0 (50.4) <0.001a WM global FAh, mean (SD) 0.33 (0.02) 0.32 (0.02) 0.007a WM global MDh, mean (SD), x10-3mm2/s 0.88 (0.04) 0.91 (0.04) 0.008a

Abbreviations: FA: fractional anisotropy; GM: grey matter; IQR: interquartile range; MD: mean diffusivity; MMSE: Mini Mental State Examination; SD: standard deviation; WM: white matter; WMH: WM hyperintensity a age and sex adjusted using ANCOVA. b age and sex adjusted using logistic regression. c 1 participant was excluded because of missing cognitive data. d 4 participants were excluded because of missing values on baseline gait speed. e Brain volumes are represented normalized to the total intracranial volume. f 1 participant was excluded because of imaging artefacts g Respectively 3 (in group included in analysis) and 1 participant(s)(in group not included in analysis) were excluded because of missing values of baseline microbleeds. h Respectively 2 (in group included in analysis) and 1 participant(s) (in group not included in analysis) were excluded because of baseline DTI artefacts.

50 BASELINE CEREBRAL SMALL VESSEL DISEASE AND GAIT DECLINE.

Results Characteristics of the study population are shown in Table 1. Mean age of the study population at baseline was 63.3 years (SD8.4) and mean follow-up duration was 5.4 years (SD0.2). Those 1 who were excluded were older, had a slower gait, had smaller GM and WM volumes, higher WMH volume, higher presence of lacunes and lower FA and higher MD parameters at baseline in comparison to those who participated (Table 1). Out of 310 participants, 48 (15.5%) developed gait impairment during follow-up, 18 had 2 already an impaired gait at baseline and 244 participants maintained a gait speed above 1.0 m/s at follow-up. In total, 11.6% showed no gait decline and 71.9% had gait decline of ≥0.1m/s. After 5 years of follow-up, there was a significant reduction in gait speed, stride length and cadence in the total study population (Table 2). 3 There were no significant associations between the baseline traditional CSVD markers (WMH volume, WM and GM volume and the number of lacunes and microbleeds) and DTI measures of the WM and changes in gait parameters (including gait speed, stride length and cadence) (Table 3) and incident gait impairment (Table 4). 4 The TBSS analysis revealed higher MD values in multiple WM tracts in participants with incident gait impairment compared to those without (Figure 2). However, these differences were not significant after additional adjustment for conventional CSVD markers. For FA values no significant differences were found between both groups (data not shown). 5 In sub-analyses, a possible threshold effect was seen for WMH volume; participants with the highest quartile of WMH volume (>11.6mL) had an increased 5-year risk for the development of gait impairment (OR 2.8, 95%CI: 1.1-7.1, p=0.03 in comparison to participants with the 1st-3rd quartiles of WMH volume, range 0.6-11.6mL), although this remained not significant 6 after correction for multiple testing (data not shown). No threshold effects were found for the other imaging measures. The region-specific sub-analyses of WMH, showed that baseline WMH volume in the infratentorial region (brainstem and cerebellum) was associated with gait decline (gait 7 speed decline β=-0.22, p=0.008; stride length decline β=-0.18, p=0.03; cadence decline β=- 0.24, p=0.01) (Table 5) and incident gait impairment after 5 years of follow-up (OR 1.8 per SD increase, 95%CI: 1.1-2.9, p=0.02) (Table 6), although this was not significant after correction 8 for multiple testing.

51 CHAPTER 3 a 0.81 0.011 -value <0.001 p n=18 0.92 (0.08) 0.78 (0.23) 1.11 (0.12) 0.92 (0.23) 100.5 (5.8) 99.7 (10.2) impairment impairment Baseline gait Baseline gait a -value <0.001 <0.001 <0.001 p n=48 1.22 (0.17) 0.89 (0.08) 1.33 (0.16) 1.03 (0.11) 110.6 (8.2) 104.9 (8.8) impairment Incident gait Incident gait a 0.64 -value <0.001 <0.001 p

n=244 No gait No gait 1.43 (0.18) 1.26 (0.15) 1.50 (0.16) 1.32 (0.13) 115.3 (8.9) 115.1 (8.4) impairment impairment a 0.01 -value <0.001 <0.001 p n=310 1.37 (0.22) 1.18 (0.22) 1.45 (0.19) 1.25 (0.19) 113.7 (9.4) 112.6 (9.8) population Total study study Total -value for difference between baseline and follow-up gait parameters calculated with a paired t-test. paired with a calculated parameters gait follow-up baseline and between difference for -value p GAITRite parameters GAITRite Gait speed (m/s) Baseline Follow-up length (m) Stride Baseline Follow-up (steps/min) Cadence Baseline Follow-up Table 2. Comparison of GAITRite parameters at baseline and follow-up baseline and at parameters GAITRite of 2. Comparison Table deviation). (standard mean represent Data a

52 BASELINE CEREBRAL SMALL VESSEL DISEASE AND GAIT DECLINE.

Table 3. Association between baseline imaging measures and changes in gait

Change in gait parameters 1 Baseline imaging characteristics Δ Gait Δ Stride Δ Cadence (n=310) speed (m/s) length (m) (steps/min) WMH volumea, per SDb -0.04 -0.09 0.03 Lacunes, per numberb 0.01 0.01 -0.02 Microbleedsc, per numberb 0.04 0.03 0.06 2 WM volume, per SDb 0.10 0.09 0.07 GM volume, per SDd 0.07 0.09 -0.02 WM global FAe, per SDb,d -0.02 -0.05 0.04 WM global MDe, per SDb,d 0.05 0.04 0.01 3 Abbreviations: FA: fractional anisotropy; GM: grey matter; MD: mean diffusivity (x10-4 mm2/s); SD: standard deviation; WM: white matter; WMH: WM hyperintensity Data are standardized beta-values. All covariates are adjusted for time between baseline and follow-up assessment and the following baseline 4 covariates: age, sex, height, gait parameters and cognitive index. a log transformed. b adjusted in addition for grey matter volume. c 3 participants were excluded because of missing values of microbleeds at baseline. d adjusted in addition for traditional CSVD markers (WMH volume, number of lacunes and microbleeds and WM 5 volume). e 2 participants were excluded for DTI analyses because of baseline DTI artefacts. *p<0.05. 6

53 CHAPTER 3

Table 4. Association between baseline imaging measures and the risk of incident gait impairment at follow-up

Baseline imaging characteristics Odds ratio (95% CI) for p-value (n=292)a incident gait impairmentb (n=48) WMH volume, per SD 1.35 (0.93-1.96)c 0.12 Lacunes, presence 0.90 (0.33-2.48)c 0.84 Microbleeds, presenced 1.55 (0.57-4.24)c 0.39 WM volume, per SD 0.98 (0.63-1.51)c 0.92 GM volume, per SD 0.89 (0.52-1.53)e 0.68 WM global FA, per SDf 0.98 (0.58-1.23)c,e 0.95 WM global MD, per SDf 0.98 (0.51-1.88)c,e 0.94

Abbreviations: CI: confidence interval; FA: fractional anisotropy; GM: grey matter; MD: mean diffusivity (x10- 4mm2/s); per SD: odds ratios per standard deviation difference from the mean; WM: white matter; WMH: WM hyperintensity. All covariates are adjusted for time between baseline and follow-up assessment and the following baseline covariates: age, sex, height, gait speed and cognitive index. a 18 participants with baseline gait speed impairment were excluded from this analysis. b defined as a gait speed <1.0 m/s at follow-up. c adjusted in addition for grey matter volume. d 3 participants were excluded because of missing values of microbleeds at baseline. e adjusted in addition for traditional CSVD markers (WMH volume, number of lacunes and microbleeds and WM volume). f 2 participants were excluded for DTI analyses because of baseline DTI artefacts.

54 BASELINE CEREBRAL SMALL VESSEL DISEASE AND GAIT DECLINE.

Discussion In this cohort study with older adults with CSVD we found no significant associations between baseline imaging markers of CSVD and gait decline or incident gait impairment 1 after 5 years, even though our population showed a mean gait decline of 0.2m/s in 5 years and a considerable amount of participants developed gait impairment (15.5%). In our TBSS analysis, we found higher baseline MD values in multiple WM tracts in participants with incident gait impairment compared to those without. However, this remained not significant 2 after additional adjustment for traditional CSVD markers. In sub-analyses, we found that participants with the highest quartile of baseline WMH volume had an increased 5-year risk of incident gait impairment. Furthermore, region-specific analyses revealed that WMH in the infratentorial region were associated with gait decline and incident gait impairment 3 after 5 years. Although, results of these sub-analyses were not significant after correction for multiple testing. Major strengths of our study include the single-centre design, the quantitative measurement of gait, the inclusion of multiple imaging markers of CSVD, including DTI measures, and 4 the follow-up duration of 5 years. Furthermore, all imaging data were analyzed by raters blinded to clinical information and adjustments for several confounders, including cognitive performance, were made. No adjustments were made for cardiovascular risk factors, as we considered them part of the causal chain of CSVD. 5 A methodological consideration includes the occurrence of attrition bias, because a considerable number of participants could not be included in the present study. As these participants were older, more disabled and had a higher load of CSVD, it is possible that the strength of the associations may have been underestimated. 6 Previous performed studies on baseline CSVD and gait decline over time are limited and their results are conflicting.87-90 These studies are mostly performed in ageing populations (mean age>72years), with often already a low gait speed (<1.0m/s) at baseline,89, 90 which make comparison to our study difficult. Furthermore, no corrections were made for multiple 7 testing in these previous studies. Extending our previous findings in which we showed that baseline CSVD is associated with incident parkinsonism, with lower body symptoms, including gait difficulties, being the dominant feature of these patients,97 we hypothesized 8 that baseline CSVD might also be associated with gait decline over time. Surprisingly, we found no associations between baseline CSVD markers and gait decline after 5 years of follow- up, despite the observation of a considerable deterioration of gait in our participants. Several possible explanations could be proposed for finding. First, CSVD is just one of multiple risk 9 factors of gait impairment, as gait is the result of a complex interaction between many (organ) systems, including the peripheral and central nerve system, cardiovascular and pulmonary system and musculoskeletal system.15 This is in accordance with the results of a recent study, showing that a high disease burden across multiple organ systems at baseline was A associated with gait decline in a community-based population aged ≥65 years.102 No specific system was found to be primarily responsible for the observed gait decline over 6 years.102

55 CHAPTER 3

Table 5. Association between baseline WMH volume per location and changes in gait

Change in gait parameters Baseline WMH volume per location Δ Gait Δ Stride Δ Cadence (n=310) speed (m/s) length (m) (steps/min) Frontal lobe WMH volumea -0.06 -0.09 0.15 Parietal lobe WMH volumea 0.01 0.05 -0.02 Temporal lobe WMH volumea -0.04 -0.03 -0.10 Occipital lobe WMH volumea -0.05 -0.05 -0.06 Sublobar WMH volumea 0.22 0.18 0.17 Limbic WMH volumea -0.01 0.03 -0.03 Infratentorial WMH volumea -0.22b -0.18b -0.24b

Abbreviations: WMH: white matter hyperintensity Data are standardized beta-values. All covariates are adjusted for time between baseline and follow-up assessment and the following baseline covariates: age, sex, height, gait parameters, cognitive index, grey matter volume and total WMH volume. a log transformed. bp<0.05. Bold values indicate significance after Bonferroni correction p( <0.007).

Table 6. Association between baseline WMH volume per location and the risk of incident gait impairment at follow-up

Baseline WMH volume per location Odds ratio (95% CI) for p-value (n=292)a incident gait impairmentb (n=48) Frontal lobe WMH volume, per SD 0.34 (0.09-1.22) 0.34 Parietal lobe WMH volume, per SD 0.75 (0.34-1.64) 0.47 Temporal lobe WMH volume, per SD 1.49 (0.73-3.03) 0.27 Occipital lobe WMH volume, per SD 1.20 (0.74-1.96) 0.46 Sublobar lobe WMH volume, per SD 0.63 (0.28-1.45) 0.63 Limbic lobe WMH volume, per SD 1.13 (0.42-3.07) 0.81 Infratentorial WMH volume, per SD 1.77 (1.10-2.85) 0.02

Abbreviations: CI: confidence interval; per SD: odds ratios per standard deviation difference from the mean; WMH: white matter hyperintensity. All covariates are adjusted for time between baseline and follow-up assessment and the following baseline covariates: age, sex, height, gait speed, cognitive index, grey matter volume and total WMH volume. a 18 participants with baseline gait speed impairment were excluded from this analysis. b defined as a gait speed <1.0 m/s at follow-up. Bold values indicate significance after Bonferroni correction (p<0.007).

56 BASELINE CEREBRAL SMALL VESSEL DISEASE AND GAIT DECLINE.

Model 1 1 1

R R 3

Model 2 2 4

p<0.01 5

p<0.05 R R

6 zz = = 19 19 yy = 2222 x x= = -13 -13

Figure 2. Differences in baseline mean diffusivity values between participants with and without incident gait impairment after 5 years 7 Voxel-wise analysis of the differences in baseline mean diffusivity (MD) values between participants with incident gait impairment (n=48) and without gait impairment (n=240; 4 participants were additionally excluded because of missing values of microbleeds and DTI artefacts). Adjusted for follow-up duration, baseline age, sex, height, gait speed, cognitive index and normalized total brain volume (model 1) and additionally for traditional CSVD markers (WMH volume, number of lacunes and microbleeds) (model 2), performed with a two sample 8 t-test, thresholded at p<0.05 and corrected for multiple comparisons. These images are superimposed onto the spatially normalized Montreal Neurological Institute (MNI) stereotactic space FA map. R indicates right side. The x, y and z coordinates represent the MNI coordinates of each slide. 9

57 CHAPTER 3

This indicates that accumulation of pathology in multiple organ systems, which in part may share the same common pathway by cardiovascular risk factors, might be a better predictor of gait decline. This might possibly also explain our result that only participants with the highest quartile of WMH at baseline seemed to have an increased risk for the development of gait impairment, as a high WMH volume might be a reflection of increased (vascular) damage to cerebral networks and possibly also to other organ systems. However, a note of caution is needed here, due to wide confidence intervals and multiple testing. Second, by initially analyzing total burden of the different CSVD markers, we could have missed region-specific associations. We therefore performed region-specific sub-analysis revealing associations between WMH in the infratentorial region and gait decline, although these associations were not significant after correction for multiple testing. Our results are in line with a cross- sectional study, which also showed that participants with WMH in the brainstem walked slower.103 An explanation for this finding might be that these WMH could damage motor fibres in the corticospinal and spinocerebellar tracts, as well as numerous cerebellovestibular connections, which are centered in a relatively small area in comparison to supratentorial regions.103 Third, it may be that progression of CSVD is associated with gait decline, rather than baseline burden of CSVD. Most of our participants had only mild to moderate severe CSVD at baseline, which may have prevented us for finding significant associations. A recent study showed that WM atrophy and WMH progression were associated with gait decline after 2.5 years of follow-up.104 A future study of the RUN DMC is underway to further investigate this. Our study is unique in using baseline DTI measures in relation to gait decline. Nevertheless, our TBSS analysis revealed no significant differences in baseline WM microstructural integrity between participants with and without incident gait impairment independent of conventional MRI markers of CSVD. DTI might however be of interest for future research as a recent study showed that change in DTI measures could serve as a sensitive marker for CSVD progression, especially MD.105 Therefore, future studies should focus on changes in DTI measures in relation to gait decline, in addition to changes in conventional MRI markers of CSVD. We hypothesize that changes in diffusion measures might be a better marker of gait deterioration than traditional CSVD markers, as loss of WM microstructural integrity might possibly underlie and precede the earlier observed relation between WM atrophy and WMH progression and gait decline.104 In conclusion, in older adults with CSVD traditional CSVD markers and WM microstructural integrity at baseline are not associated with gait decline or incident gait impairment after 5 years. This result might, however, in part be driven by the attrition bias in our study, despite the fact that a high percentage of our participants experienced a significant gait decline. Future studies should be directed at changes in these cerebral imaging markers in relation to gait decline, as this could provide more insight into the role of (progression of) CSVD to gait deterioration, which more and more burdens the health care system of aging societies.

58 BASELINE CEREBRAL SMALL VESSEL DISEASE AND GAIT DECLINE.

4. White matter changes and gait decline

Submitted as: H.M. van der Holst*, A.M. Tuladhar*, V. Zerbi, I.W.M. van Uden, K.F. de Laat, E.M.C. van Leijsen, M. Ghafoorian, B. Platel, M.I. Bergkamp, A.G.W. van Norden, D.G. Norris, E.J. van Dijk, A.J. Kiliaan, F-E de Leeuw.

White matter changes are associated with gait decline in cerebral small vessel disease

*Both authors contributed equally. CHAPTER 4

Abstract

Objective: To investigate the longitudinal associations between (micro)structural brain changes and gait decline in cerebral small vessel disease using diffusion tensor imaging. Methods: From the Radboud University Nijmegen Diffusion tensor and Magnetic resonance imaging Cohort (RUN DMC), a prospective cohort of participants with cerebral small vessel disease aged 50-85 years, 275 participants were included. Gait (using GAITRite) and magnetic resonance imaging measures were assessed during baseline (2006-2007) and follow-up (2011-2012). Linear regression analysis was used to investigate the association between changes in conventional magnetic resonance and diffusion tensor imaging measures and gait decline. Tract-based spatial statistics analysis was used to investigate region-specific associations between changes in white matter integrity and gait decline. Results: Of the 275 participants, 56.2% were male, mean age was 62.9 years (SD8.2), mean follow-up duration was 5.4 years (SD0.2) and mean gait speed decline was 0.2m/s (SD0.2). Stride length decline was associated with white matter atrophy (β=0.16,FDR- adjusted p=0.04) and increase in mean white matter radial diffusivity β( =-0.14, FDR- adjusted p=0.04), independent of age, sex, height, follow-up duration and baseline stride length. Tract-based spatial statistics analysis showed significant associations between stride length decline and fractional anisotropy decrease and mean diffusivity increase (primarily explained by radial diffusivity increase) in multiple white matter tracts, with the strongest associations found in the corpus callosum and corona radiata, independent of traditional cerebral small vessel disease markers. Conclusion: white matter atrophy and loss of white matter integrity are associated with gait decline in older adults with cerebral small vessel disease after 5 years of follow-up.

62 WHITE MATTER CHANGES AND GAIT DECLINE.

Introduction Gait disturbances are prevalent in older adults aged ≥65 years and have important 85, consequences as they can lead to falls, functional dependence and institutionalization. 1 86 Cerebral small vessel disease (CSVD) ranks highest among vascular causes of gait decline, with evidence mainly coming from cross-sectional studies.17, 106 A few studies have investigated whether baseline CSVD can predict gait decline over time. However, their results are inconclusive.87-89 In a previous study, we found no associations between baseline 2 CSVD and gait decline after 5 years in our CSVD population.107 One possible hypothesis is that progression of CSVD, rather than baseline CSVD load, is associated with gait decline over time. The results of recent longitudinal population-based studies investigating the association between progression of CSVD and gait decline are however conflicting.90, 104, 3 108 Moreover, these studies did not include the whole CSVD spectrum, nor was the white matter (WM) integrity, which can be assessed using diffusion tensor imaging (DTI), taken into account. Since a recent study showed that scalar measures of DTI were sensitive markers of CSVD progression,105 we hypothesized that changes in DTI measures are stronger associated 4 with gait decline than progression of the traditional CSVD markers. Here, we investigated the longitudinal associations between changes in the traditional CSVD markers (WMH, lacunes, microbleeds and brain atrophy) and changes in WM integrity (assessed by DTI), and gait decline in a population of adults with CSVD, aged 50-85 years, over 5 a period of 5 years.

63 CHAPTER 4

Methods Study population The Radboud University Nijmegen Diffusion tensor and Magnetic resonance Cohort study (RUN DMC study) prospectively investigates the risk factors and clinical consequences of brain changes as assessed by MRI. This cohort study consists of 503 participants with CSVD, aged 50-85 years at baseline (2006). The recruitment, study rationale and protocol of the RUN DMC study have been described elsewhere.47 Inclusion criteria were age 50-85 years and CSVD on neuroimaging (defined as the presence of WMH or lacunes of presumed vascular origin).48 All consecutive patients referred to our outpatient clinic who underwent diagnostic brain imaging for several reasons (e.g. stroke, TIA, cognitive complaints) were eligible for participation. Main exclusion criteria were: parkinsonism, dementia, life expectancy < 6 months, non-CSVD related WM lesions and MRI contra-indications.47 Follow-up assessment was completed in 2012. Of the 503 baseline participants, 398 participated in the follow-up assessment. For the present study, 123 participants were additionally excluded, yielding a final sample of 275. Exclusion reasons at baseline and follow- up were: missing data on MRI and gait data, territorial infarcts, parkinsonism and conditions associated with gait impairment other than CSVD and parkinsonism (see flowchart Figure 1).

Standard protocol approvals, registration, and patient consents All participants signed an informed consent form. The Medical Review Ethics Committee region Arnhem-Nijmegen approved the study.

Gait measurement and gait impairment The assessment of gait was performed by using a 5.6 meter electronic portable walkway (GAITRite, MAP/CIR Inc., Havertown, PA), which has an excellent test-retest reliability and validity.93, 94 Participants were instructed to walk over the walkway at their comfortable walking speed. In order to measure steady-state walking, they started two meters before the walkway and stopped two meters behind it. The following gait parameters were averaged over two walks: gait speed (m/s) and its components stride length (m) (the distance between the heel points of two consecutive footprints of the same foot) and cadence (number of steps per minute). Changes in these gait parameters were calculated as the difference between follow-up and baseline assessment.

64 WHITE MATTER CHANGES AND GAIT DECLINE.

Baseline study population n = 503 1

Lost to follow-up n = 2 2 Deceased n = 49

Unable to visit research centre n = 54 • illness that prevented visit (19) • moved (5) • lack of time (30) 3 Follow-up study population n = 398

Exclusion criteria n = 123 4 • MRI contra-indications/artefacts/missing (46) • Missing gait data at follow-up (12) • Territorial infarcts baseline and follow-up (43) • Conditions associated with gait impairment other than CSVD at baseline and follow-up (13) • Parkinsonism at follow-up (6) 5 • DTI artefacts (3)

Participants for analysis n = 275 6 Figure 1. Flowchart of the study sample Abbreviations: DTI: diffusion tensor imaging;MR I: magnetic resonance imaging; CSVD: cerebral small vessel disease. Of the 503 baseline participants, 2 participants were lost to follow-up, 49 had died and 54 refused an in- 7 person follow-up examination, but their clinical endpoints were available; 398 participated in the follow-up assessment. For the present study, we included 275 participants, 123 participants were additionally excluded because of because of (i) MRI contra-indications, MRI artefacts or missing values at follow-up (n=46), (ii) missing data on follow-up GAITRite (n=12) (because they were wheelchair bound, because of home visit or because of 8 technical problems), (iii) territorial infarcts at baseline and follow-up imaging (n=43), because these infarcts were considered as a potential confounder, (iv) conditions associated with gait impairment other than CSVD which prevented participants from walking unaided at baseline and follow-up (n=13) (joint fusion, severe arthritis, severe polyneuropathy, leg amputation, severe vision problems, severe cardiopulmonary diseases, severe peripheral arterial disease and psychogenic gait disturbance), (v) parkinsonism during follow-up examination 9 (n=6), because apart from CSVD other pathologies as amyloid pathology, Lewy body pathology and nigrastriatal dopaminergic loss could play a role in gait deterioration in these patients, and (vi) DTI artefacts (n=3).

65 CHAPTER 4

MRI protocol A cerebral MRI was acquired on a 1.5-Tesla scanner at baseline and follow-up (baseline: Magnetom Sonata; follow-up: Magnetom Avanto Tim (76x32); Siemens Medical Solutions, Erlangen, Germany). The same 8-channel head coil was used at baseline and follow- up. The protocol included: a T1-weighted 3D magnetization-prepared rapid gradient- echo (MPRAGE) imaging (baseline: repetition time (TR)/ echo time (TE)/ inversion time (TI) 2250ms/3.68ms/850ms, flip angle=15°, voxelsize 1.0x1.0x1.0mm; follow-up: TR/TE/ TI 2250ms/2.95ms/850ms, flip angle=15°, voxelsize 1.0x1.0x1.0mm); a Fluid-attenuated inversion recovery (FLAIR) sequence (baseline: TR/TE/TI 9000ms/84ms/2200ms, voxelsize 1.2x1.0x5.0mm (interslice gap 1mm); follow-up: TR/TE/TI 14240ms/89ms/2200ms, voxelsize 1.2x1.0x2.5mm (interslice gap 0.5mm); a transversal T2*weighted gradient echo sequence (baseline and follow-up: TR/TE 800ms/26ms, voxelsize 1.3x1.0x5.0mm (interslice gap 1mm)) and a DTI sequence (baseline: TR/TE 10100ms/93ms, voxelsize 2.5x2.5x2.5mm; 4 unweighted scans, 30 diffusion weighted scans with b-value=900 s/mm²; follow-up: TR/TE 10200ms/95ms, voxelsize 2.5x2.5x2.5mm; 7 unweighted scans, 61 diffusion weighted scans with b-value=900 s/mm²).

Traditional CSVD markers and brain volumetry Traditional CSVD makers (WMH, lacunes and microbleeds) were rated according to the STRIVE criteria.5 Lacunes were manually rated on FLAIR/T1-weighted scans and microbleeds on T2*-weighted MRI scans by raters blinded to clinical data. The follow-up FLAIR images were resliced to match the slice thickness of baseline FLAIR images, limiting the differences in partial volume effects between baseline and follow-up scans. Intrarater and interrater reliabilities were good (for lacunes: weighted kappa values 0.87 and 0.95, respectively, and for microbleeds: 0.85 and 0.86, respectively). WMH were segmented by using an in-house developed semi-automatic detection method on baseline and follow-up FLAIR sequences.109 All scans were visually checked by 1 rater and corrections were made when segmentation failures had occurred. Total WMH volume was calculated by summing all segmented areas multiplied by slice thickness. Automated segmentation on T1 images of baseline and follow-up was performed using Statistical Parametric Mapping 12 unified segmentation routines (SPM12; Wellcome Department of Cognitive Neurology, University College London, United Kingdom, http:// www.fil.ion.ucl.ac.uk/spm/software/spm12/), in order to obtain grey matter (GM), WM and cerebrospinal fluid (CSF) probability maps. To avoid the erroneous segmentation of WM regions with WMH as GM, the T1 images were first corrected using the binary maps of WMH by replacing the voxel intensities of WMH with the average intensity of the normal-appearing WM on the T1 images. The volumes were calculated by summing all the voxel volumes belonging to the tissue class. All images were visually checked for co-registration errors and motion and/or segmentation artifacts. Total brain volume was taken as the sum of total GM

66 WHITE MATTER CHANGES AND GAIT DECLINE. and WM volume. GM volume was composed of the volume of the neocortex, basal ganglia and thalamus. To account for inter-scan-effects, we corrected the normalized follow-up brain volumes 1 for the difference in intracranial volume (ICV; sum of GM, WM and CSF) between baseline and follow-up by multiplying all volumes by the factor ‘ICV baseline/ICV follow-up’. Next, all volumes were normalized to the baseline ICV to adjust for head size.98 We calculated brain volume change and changes in the number of lacunes and microbleeds as the difference 2 between follow-up and baseline.

DTI analysis Diffusion data were preprocessed and analyzed according to a previous described 3 procedure47 for baseline and follow-up DTI scans. In short, after eddy current and motion artifacts corrections on the raw diffusion data, we created fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD) and radial diffusivity (RD) images using DTIFit within FSL, which were then fed into the tract-based spatial statistics (TBSS) pipeline.46 An FA 4 skeleton was created by thinning the mean FA image based on the FA values. This skeleton was then thresholded at 0.3 to include major WM tracts. The images of MD, AD and RD were subsequently projected on this mean FA skeleton, by applying the projection vectors from each participant’s FA-to-skeleton transformation. Changes in the diffusion parameters were 5 derived by calculating the difference between DTI measures of the skeleton at follow-up and baseline.

Statistical analysis 6 Statistical analyses were performed using IBM (Armonk, NY) SPSS Statistics 20. To compare the baseline characteristics of participants who were included and those not, we used age and sex-adjusted ANCOVA or logistic regression analysis. For those included, gait and imaging characteristics at baseline and follow-up were compared by using a paired t-test, Wilcoxon 7 signed rank test or McNemar test when appropriate. Multiple linear regression analysis was used to investigate the association between change in each gait variable and change in the different MRI and DTI measures. Adjustments were made for age, sex, follow-up duration 8 (time between baseline and follow-up assessment), height and baseline gait variable. WMH volume was log transformed, because of the skewed distribution. The variance inflation factor (VIF) was calculated for all regression models to test for the presence of multicollinearity. The VIF scores were low for all models (scores were below 3, where VIF-scores >5 are considered 9 to reflect high multicollinearity). Regression coefficients were presented as standardized beta-values. False discovery rate (FDR)-correction was used at a q-value<0.05 to correct for multiple comparisons.110 Voxel-wise statistical analyses for changes in TBSS data and individual gait variables were A performed by using permutation-based statistical interference tool for non-parametric approach as part of the FSL toolbox (randomize). The number of permutation tests was

67 CHAPTER 4 set at 5000. Significant associations were determined by using a threshold-free cluster enhancement with a p-value<0.05, corrected for multiple comparisons. Adjustments were made for follow-up duration, age, sex, height and baseline gait variable (model A) and additionally for changes in MRI measures (including WMH volume, number of lacunes and microbleeds, WM and GM volume) (model B).

Table 1. Baseline characteristics of the study population

Baseline characteristics Participants Participants not p-value for included included difference Demographics n=275 n=228 Age, mean (SD), years 62.9 (8.2) 69.0 (8.3) <0.001a Male sex, No. (%) 155 (56.4) 129 (56.6) 0.96b MMSE score, mean (SD) 28.5 (1.5) 27.7 (1.8) 0.004a Gait characteristics n=275 n=224c Gait speed, mean (SD), m/s 1.4 (0.2) 1.2 (0.3) <0.001a Gait impairment (<1.0m/s), No. (%) 12 (4.4) 58 (25.9) <0.001b Imaging measuresd n=275 n=227e WMH volume, median (IQR), mL 2.2 (0.8-6.3) 7.2 (2.6-15.9) 0.001a WM volume, mean (SD), mL 467.2 (37.8) 439.6 (50.2) <0.001a GM volume, mean (SD), mL 622.3 (49.0) 586.8 (50.2) <0.001a Lacunes presence, No. (%) 44 (16.0) 88 (38.6) <0.001b Microbleeds presence f, No. (%) 39 (14.3) 44 (19.5) 0.81b WM global FA, mean (SD) 0.33 (0.02) 0.32 (0.02)g 0.01a WM global MD, mean (SD), x10-3 mm2/s 0.88 (0.04) 0.91 (0.05)g <0.001a

Abbreviations: FA: fractional anisotropy; GM: grey matter; IQR: interquartile range; MD: mean diffusivity;MMSE : Mini Mental State Examination; SD: standard deviation; WM: white matter; WMH: white matter hyperintensity a age and sex adjusted using ANCOVA. b age and sex adjusted using logistic regression. c 4 participants were excluded because of missing values on baseline gait. d Brain volumes are represented normalized to the total intracranial volume. e 1 participant was excluded because of imaging artefacts. f 2 participants in both groups were excluded because of missing values of baseline microbleeds. g 3 participants were excluded because of baseline DTI artefacts.

68 WHITE MATTER CHANGES AND GAIT DECLINE.

Results The total study population consisted of 275 participants with a mean (SD) follow-up duration of 5.4 (0.2) years and a mean age at baseline of 62.9 (8.2) years; 56.4% was male. Characteristics 1 of the participants included in this study and those not included are shown in Table 1. Those not included were older, had a poorer cognitive performance, slower gait speed, higher WMH volume, more lacunes, lower WM and GM volume and lower FA and higher MD values of the WM. 2 Table 2 shows the gait and imaging characteristics of our study population at baseline and follow-up. Mean gait speed decline was 0.2m/s (SD0.2) over 5 years (p<0.001, one sample t-test), mainly due to reduction in stride length (mean decline of 0.2m (SD0.1), p<0.001). We found a non-significant decrease of cadence (mean decline 0.9 steps/min (SD7.5), p=0.06). 3 There was a significant increase of WMH volume, presence of lacunes and microbleeds and MD, RD and AD values of WM tracts between baseline and follow-up. A significant decrease was seen for WM and GM volume and FA value of the WM tracts during follow-up (Table 2). 4 Table 2. Comparison of gait and imaging measures at baseline and follow-up (n=275)

Baseline Follow-up p-value Gait characteristics Gait speed, mean (SD), m/s 1.38 (0.21) 1.19 (0.21) <0.001a 5 Stride length, mean (SD), m 1.46 (0.18) 1.26 (0.18) <0.001a Cadence, mean (SD), steps/min 113.9 (9.4) 113.0 (9.3) 0.06a Imaging measures WMH volume, median (IQR), mL 2.2 (0.8-6.3) 2.8 (1.2-8.0) <0.001b 6 WM volume, mean (SD), mL 467.6 (37.9) 457.0 (42.8) <0.001a GM volume, mean (SD), mL 622.8 (49.1) 612.0 (50.2) <0.001a Lacunes presence, No. (%) 44 (16.0) 61 (22.2) <0.001c Microbleeds presence d, No. (%) 39 (14.3) 56 (20.4) <0.001c 7 FA of skeleton, mean (SD) 0.49 (0.03) 0.47 (0.03) <0.001a MD of skeleton, mean (SD), x10-3 mm2/s 0.80 (0.04) 0.82 (0.05) <0.001a

Abbreviations: AD: axial diffusivity; FA: fractional anisotropy; GM: grey matter; IQR: interquartile range; MD: mean diffusivity;RD : radial diffusivity; SD: standard deviation; WM: white matter; WMH: white matter hyperintensity 8 a paired T-test b Wilcoxon signed rank test c McNemar test d respectively 2 and 1 participant(s) had missing values of microbleeds at baseline and follow-up 9

69 CHAPTER 4

Progression of traditional CSVD markers and gait decline A decline in WM volume was associated with a decline in stride length (β=0.16, FDR-adjusted p=0.041), after adjustment for age, sex, follow-up duration, height, and baseline stride length (Table 3). No significant associations were found for changes in stride length and the other CSVD markers, or for changes in gait speed or cadence and changes in all traditional CSVD markers.

Table 3. Association between changes in imaging measures and changes in gait

Change in gait parameters Change in imaging measures Δ Gait speed Δ Stride length Δ Cadence (m/s) (m) (steps/min) ΔWMH volume a, mL -.03 -.05 .02 Δ WM volume, mL .10 .16 .04 Δ GM volume, mL .08 .10 .06 Δ Lacunes, No. .04 .01 .06 Δ Microbleeds, No. -.09 -.10 -.07 Δ FA of skeleton .06 .10 .01 Δ MD of skeleton, x10-4 mm2/s -.06 -.12 .05 Δ RD of skeleton, x10-4 mm2/s -.07 -.14 .02 Δ AD of skeleton, x10-4 mm2/s .01 -.03 .07

Abbreviations: AD: axial diffusivity; FA: fractional anisotropy; GM: grey matter; MD: mean diffusivity; RD: radial diffusivity;WM : white matter; WMH: white matter hyperintensity. Data are standardized beta-values. Adjustments were made for age, sex, follow-up duration, height and baseline gait parameter (baseline gait speed, stride length or cadence respectively). Bold values indicate significance at p<0.05, false discovery rate-corrected. Δ indicates difference between follow-up and baseline assessment. a log transformed.

Changes in DTI parameters and gait decline Only significant associations were found for changes in DTI measures and changes in stride length, and not for changes in gait speed or cadence .An increase of RD value of WM tracts was associated with a decline in stride length (β= -0.14, FDR-adjusted p=0.041). Marginally significant associations were found between an increase of MD and stride length decline (β=- 0.12, FDR-adjusted p=0.054) (Table 3). Our longitudinal TBSS analysis showed significant associations between an increase of MD (primarily explained by an increase of RD) and to a lesser extent decrease of FA and decline in stride length (Figure 2), independent of traditional CSVD markers. These associations were primarily found in the corpus callosum, and superior and posterior corona radiata. In contrast, no significant associations were found between change in AD and decline in stride length or between changes in DTI measures and decline in gait speed or cadence after additionally controlling for the traditional CSVD markers (data not shown).

70 WHITE MATTER CHANGES AND GAIT DECLINE. <0.05 and p 1

A Figure 2. Association between decrease in stride length and changes in diffusion tensor imaging measures after 5 years of follow-up of years 5 after imaging measures tensor in diffusion in stride length and changes decrease between Association 2. Figure Voxel-wise analysis of relation between changes in stride length (in centimetres) and changes in different diffusion tensor imaging measures, thresholded at thresholded measures, tensor imaging diffusion in different and changes length (in centimetres) in stride changes between analysis of relation Voxel-wise in changes length (model 1) and additionally for stride height and baseline duration, sex, follow-up age, made for were Adjustments multiple comparisons. for corrected (model 2). These images volume) matter and white matter and grey number of lacunes, microbleeds hyperintensities, matter (white characteristics disease small vessel cerebral the represent and z coordinates The x, y, map. anisotropy fractional space (MNI) stereotactic Institute Neurological Montreal normalized the spatially superimposed onto are slide. of each MNI coordinates tracts matter white values of the significant radial diffusivity diffusivity and in mean changes between regression) (linear the relation shows the image to next The scatterplots respectively. length (in centimetres), in stride analysis and changes (TBSS) statistic spatial in tract-based found

71 CHAPTER 4

Discussion In this longitudinal study, we found that WM atrophy and loss of WM integrity (indicated by an increase of MD and RD, and to a lesser extent decrease of FA) were associated with gait decline by affecting stride length in older adults with CSVD after 5 years of follow-up. Changes in DTI measures associated with stride length decline were primarily found in corpus callosum, and posterior and superior corona radiata, and were independent of traditional CSVD markers. In contrast, progression of the other CSVD markers, including increase of WMH volume, number of lacunes and microbleeds and GM atrophy, were not associated with gait decline. Main strengths of this longitudinal and single-centre study include the quantitative assessment of gait parameters on the same GAITRite at baseline and follow-up, the follow-up duration of 5.4 years, in which participants showed a significant gait decline and progression of CSVD, and the exclusion of participants with conditions associated with gait impairment other than CSVD which allowed us to further elucidate the role of CSVD in gait decline. Several methodological issues need to be addressed. First, the dropout rate (45% of the study population) may have resulted in attrition bias. However, as those who were not included had a higher load of CSVD and a slower gait at baseline it may be that the found associations have been underestimated. Second, due to the observational design of our study, causal inference cannot be reliably made. The possibility of reverse causality, indicating that gait deterioration leads to a sedentary lifestyle and thereby to progression of CSVD and reduction of microstructural integrity, cannot be excluded. Third, we cannot rule out the possibility of residual confounding by unmeasured variables. Fourth, the effect of scanner upgrade (baseline: Avanto and follow-up: Avanto MRI scanner, using the same Siemens head coil) is unknown. It has previously been found that the volumetric measures remained reliable, even after scanner upgrade and that the variance of the volume differences relative to test- retest reproducibility did not significantly change, however it may introduce a bias in the mean volume differences.111 Therefore, our results on macrostructural brain changes have to be interpreted with caution. The DTI protocol did not differ between baseline and follow- up, except from the number of diffusion weighted scans (30 versus 61 diffusion-encoding gradient directions). The effect of number of gradient directions is however limited, as we have applied the diffusion tensor model (with 6 degree of freedom). With regard to the scanner upgrade, a previous study showed that the DTI parameters did not differ between the scanners.112 Therefore, we consider the DTI results as robust. We found that a decline in gait speed is primarily caused by a decline in stride length and not in cadence in our population. This is in line with our previous finding that stride length is a more sensitive marker for gait abnormalities in CSVD compared to cadence and gait speed.17 Furthermore, we showed that WM atrophy and changes in DTI measures were associated with gait decline in older adults with CSVD. The association between WM atrophy and gait decline has been reported earlier by Callisaya et al.104 In contrast to this study, we found no significant associations between gait decline and progression of WMH volume, or progression of other traditional CSVD markers. This could possibly be explained by the limited contribution of

72 WHITE MATTER CHANGES AND GAIT DECLINE. each of these CSVD markers to gait decline, possible regional-specific associations of these CSVD markers on gait decline or a threshold-effect rather than a dose-dependent relation, as 88 some evidence has been found for WMH. Furthermore, it might be that WM microstructural 1 integrity is a moderator in the association between progression of traditional CSVD markers and gait decline, as one cross-sectional study showed that in participants with a greater WM microstructural integrity WMH were less strongly associated with gait in comparison to those with a low WM microstructural integrity.113 2 Regional analysis of DTI determined several WM tracts involved in gait decline commonly affected by CSVD pathology.106 Consistent with, and extending our previous cross-sectional study,92 we demonstrated that the strongest associations between changes of DTI measures and decline in stride length were found in corpus callosum and corona radiata. The corpus 3 callosum is an important WM tract in motor control, as this region contains commissural fibres connecting multiple cortical areas involved in gait planning, initiation and execution, including frontal, parietal and occipital cortices.114 Fibres from these regions converse into the corona radiata that as such contains projection fibres that are involved in motor pathways 4 and thus plays a pivotal role in motor function.115 Our results suggest that progression of disruption of these WM tracts is associated with gait decline in CSVD. We found the strongest association with increase in MD, and especially RD values of the WM tracts and gait decline. This is consistent with data from a recent study showing that MD 5 is a more sensitive marker for CSVD progression in comparison to FA in a CSVD population over a period of 3 years.105 An increase of MD, primarily explained by an increase in RD, is thought to represent demyelination in homogeneous parallel WM regions,116 which might be related to volume reduction of WM. This result might provide some support for the role of 6 (ischemic) demyelination in CSVD related gait decline above axonal degeneration (reflected by an increase in AD), which has also been described in neuropathological studies of CSVD.117 However, as DTI measures are dependent on eigenvalue sorting, it may be difficult to obtain reliable measures in complex WM architecture (e.g. areas with crossing fibres) or pathology 7 (e.g. CSVD)118 and therefore our results should be interpreted with caution. Our results might suggest that changes in MD of the WM, especially of the corpus callosum and corona radiata, could serve as an early marker of gait decline in an CSVD population. 8 In conclusion, our data suggest that WM atrophy and loss of WM integrity, especially of the corpus callosum and corona radiata, are associated with gait decline over a period of 5 years in older adults with CSVD. These findings favour a role for WM pathology progression in gait decline in patients with CSVD and should therefore be considered as one of the possible 9 causes of gait decline. Future studies should investigate the reproducibility of our results and the potential of DTI as surrogate and early marker of gait impairment in CSVD, for example in clinical trials. Meanwhile, clinical practitioners should focus on prevention strategies directed against progression of WM pathology in order to maintain ambulatory function in an aging A society.

5. Baseline cerebral small vessel disease and incident parkinsonism

Published as: H.M. van der Holst, I.W.M. van Uden, A.M. Tuladhar, K.F. de Laat, A.G.W. van Norden, D.G. Norris, E.J. van Dijk, R.A.J. Esselink, B. Platel, F-E de Leeuw.

Cerebral small vessel disease and incident parkinsonism: The RUN DMC study. Neurology, 2015 Nov;85(18):1569-77. CHAPTER 5

Abstract

Objective: To investigate the relation between baseline cerebral small vessel disease (CSVD) and the risk of incident parkinsonism using different MRI and diffusion tensor imaging (DTI) measures. Methods: In the Radboud University Nijmegen Diffusion tensor and Magnetic resonance Cohort (RUN DMC) study, a prospective cohort study, 503 elderly participants with CSVD and without parkinsonism were included in 2006. During follow-up (2011-2012), parkinsonism was diagnosed according to UK Brain Bank criteria. Cox regression analysis was used to investigate the association between baseline imaging measures and incident all-cause parkinsonism and vascular parkinsonism (VP). Tract-Based Spatial Statistics analysis was used to identify differences in baseline DTI measures of white matter (WM) tracts between participants with VP and without parkinsonism. Results: Follow-up was available from 501 participants (mean age 65.6 years; mean follow-up duration 5.2 years). Parkinsonism developed in 20 participants; 15 were diagnosed with VP. The 5-year risk of (any) parkinsonism was increased for those with a high white matter hyperintensity (WMH) volume (Hazard ratio (HR) 1.8 per SD increase, 95% confidence interval (CI) 1.3-2.4) and a high number of lacunes (HR 1.4 per number increase, 95% CI 1.1-1.8) at baseline. For VP, this risk was also increased by the presence of microbleeds (HR 5.7, 95% CI 1.9-16.8) and a low grey matter volume (HR 0.4 per SD increase, 95% CI 0.2-0.8). Lower fractional anisotropy values in bifrontal WM tracts involved in movement control were observed in participants with VP compared to participants without parkinsonism. Conclusions: CSVD at baseline, especially a high WMH volume and a high number of lacunes, is associated with incident parkinsonism. Our findings favour a role of CSVD in the aetiology of parkinsonism.

76 BASELINE CEREBRAL SMALL VESSEL DISEASE AND INCIDENT PARKINSONISM.

Introduction Cerebral small vessel disease (CSVD) is a frequent finding on brain imaging of the elderly 9 119 population and has been identified as a cause of motor impairment and gait and balance 1 decline over time.120 CSVD has also been related to parkinsonism, with evidence coming from cross-sectional autopsy studies that found pathological proof of CSVD in patients with parkinsonism, who did not exhibit evidence of histopathologic findings compatible with parkinsonism, including Lewy bodies or tau inclusions.20, 21 Whether parkinsonism is a direct 2 consequence of CSVD or a coincidental finding is unknown. The imaging spectrum of CSVD is rapidly expanding from lesions visible on conventional MRI, including white matter hyperintensities (WMH), lacunes, microbleeds, and (sub)cortical atrophy,5 to changes in diffusion measures of the white matter (WM) assessed by diffusion 3 tensor imaging (DTI),91 which is regarded as an index of WM structural integrity. Recent cross- sectional DTI studies have shown a relation between diffusion abnormalities in the WM and parkinsonism121, 122; however, longitudinal studies investigating the role of these MRI and DTI imaging characteristics in the development of parkinsonism are currently lacking. 4 We therefore prospectively investigated the relation between CSVD, using baseline MRI and DTI measures, including tract-based spatial statistics (TBSS), and the development of parkinsonism, in order to gain insight into the role of CSVD in incident parkinsonism. 5

77 CHAPTER 5

Materials and methods Study population This study is embedded in the Radboud University Nijmegen Diffusion tensor and Magnetic resonance Cohort (RUN DMC) study, a prospective cohort study that investigates the risk factors and clinical consequences of functional and structural brain changes as assessed by MRI in 503 independently living elderly participants with CSVD. The primary outcome of the longitudinal part of the RUN DMC study is incident parkinsonism and dementia. The recruitment, study rationale and protocol of the RUN DMC study have been described in detail elsewhere.47 CSVD was defined as the presence of any WMH or lacunes of presumed vascular origin on brain imaging,48 because the onset of CSVD is often insidious and clinically heterogeneous with acute symptoms (transient ischemic attacks (TIA) or lacunar syndromes), or subacute symptoms, including cognitive, motor or mood disturbances.123 All consecutive patients referred to our department who underwent diagnostic brain imaging (CT or MRI scan) for several reasons (e.g., stroke, TIA, cognitive complaints) were selected for participation. Inclusion criteria were age between 50-85 years and CSVD on brain imaging. Main exclusion criteria were parkinsonism, dementia, CSVD mimics, and MRI contra-indications. Patients eligible because of a lacunar syndrome were included >6 months after the event. Baseline assessment, including an extensive cognitive and motor evaluation and a cerebral MRI, took place in 2006 among 503 participants. In 2011-2012, this assessment was repeated; 2 participants were lost to follow-up (but not deceased), 49 had died and 54 refused an in- person follow-up, but their clinical endpoints were available; 398 participated in the follow- up examination (Figure 1).

Standard protocol approvals, registrations, and patient consents All participants signed an informed consent form. The Medical Review Ethics Committee region Arnhem-Nijmegen approved the study.

78 BASELINE CEREBRAL SMALL VESSEL DISEASE AND INCIDENT PARKINSONISM.

Potential participants invited by letter n = 1004 Exclusion criteria n = 172 1 Contra-indication MRI / claustrophobia n = 105 Eligible after telephone conversation n = 727 Non responders 2 n = 202 Participants n = 525

Exclusion n = 22: 14 MRI contra-indications / claustrophobia 3 1 died before MRI Baseline study population 1 multiple sclerosis n = 503 1 parkinsonism

Free of dementia and parkinsonism 4 dementia completed MRI protocol 1 language problem 4

Lost to follow-up n = 2 5 Follow-up study population n = 501

Deceased n = 49 but clinical endpoints were obtained Participants available for renewed 6 assessment n = 452 Unable to visit our research centre n = 54 but clinical endpoints were assessed Follow-up cognitive, motor and MRI 19 illness that prevented visit 7 5 removal from the area investigation 30 lack of time n = 398

Figure 1. Flowchart of RUN DMC study design of baseline and follow-up 8 Baseline and follow-up study population are presented in double lined boxes. Abbreviations: MRI: Magnetic Resonance Imaging

9 Screening for parkinsonism A flowchart of the screening for parkinsonism is shown in figure 2. The presence of parkinsonian signs was evaluated during an in-person follow-up assessment (n=398) by 2 trained residents in neurology by using the motor part of the Unified Parkinson’s Disease A Rating Scale (UPDRS-m, 27 items, score 0-108).124 Parkinsonism was defined as the presence of bradykinesia and at least one of the 3 other following signs: tremor, rigidity, or gait and

79 CHAPTER 5 postural instability, according to the UK Parkinson’s Disease Society Brain Bank criteria.125 We screened the presence of these 4 signs on the basis of previously established parkinsonian sign scores derived from the UPDRS-m,126 including limb bradykinesia (based on 8 items: right and left finger taps, handgrip, hand pronation-supination and leg agility), rigidity (based on 5 items: rigidity of neck and the 4 extremities), tremor (based on 7 items: rest tremor of lip/chin and 4 extremities and action tremor of both arms) and parkinsonian gait (based on 5 items: arise from chair, posture, gait, postural stability and body bradykinesia). We considered bradykinesia as present when ≥1 items on limb bradykinesia had a score of ≥2,18, 124 to guarantee a high sensitivity of this main symptom of parkinsonism. The other 3 signs (tremor, rigidity, and gait and postural instability) were considered present when the participant had either ≥2 items with a score of ≥1 or 1 item with a score of ≥2 in that specific category. Participants were considered screen-positive when (1) they were already diagnosed with parkinsonism by a neurologist after baseline assessment or (2) they had bradykinesia and one or more of the other 3 signs,127 according to abovementioned criteria, or (3) had an UPDRS-m score ≥10, but did not meet the criteria mentioned in (1) or (2). We added this last criterion because other co-morbidities (e.g., stroke, severe polyneuropathy, or rheumatic disease) often influence the UPDRS-m scores, which could hinder the evaluation of the presence of parkinsonism. Of the 398 participants, 68 were considered screen-positive and 40 of them were subsequently examined by a neurologist specialized in movement disorders (R.A.J.E ) for the presence of parkinsonism (10 were diagnosed with parkinsonism, 30 were not). The remaining 28 participants refused this additional evaluation, and for them a consensus diagnosis of parkinsonism was made by a panel, consisting of 2 neurologists, one of whom was specialized in movement disorders (R.A.J.E). They reviewed all available information on motor performance and imaging, including (1) UPDRS-m scores at baseline and follow- up assessment; (2) information from follow-up neurological examination, including muscle strength, gait, upper motor neuron signs, sensory deficits, (primitive) reflexes; (3) medical history; (4) medication; (5) follow-up MRI-scan, or if not available, baseline imaging (n = 12); and (6) if applicable, information on the presence of parkinsonism from their treating neurologist. Of these 28 participants, 8 were diagnosed with parkinsonism. For the participants who did not participate in person (49 deceased and 54 were not able to visit our research centre), medical records were reviewed and their treating physician was contacted for information on the presence of parkinsonism. In 2 participants, the diagnosis parkinsonism was reported; after review by the panel, these diagnoses were confirmed, yielding a total of 20 participants with incident parkinsonism. Parkinsonism was diagnosed based on the UK Parkinson’s Disease Society Brain Bank criteria for Idiopathic Parkinson’s disease (IPD),125 Zijlmans et al.21 criteria for vascular parkinsonism (VP), and National Institute of Neurological Disorders and Stroke Society for Progressive Supranuclear Palsy criteria for progressive supranuclear palsy (PSP).125 VP requires the

80 BASELINE CEREBRAL SMALL VESSEL DISEASE AND INCIDENT PARKINSONISM. presence of relevant cerebrovascular disease on neuroimaging, operationalized in our study as WMH beginning to become confluent (Fazekas score ≥2),128 or the presence of lacunes in 21 basal ganglia or thalamus. Participants with drug-induced parkinsonism were excluded (n 1 = 1). The age at onset of parkinsonism was defined as the midpoint between the date on which parkinsonism was first identified and baseline RUN DMC assessment,129 or if applicable the date at which participants were last reviewed by a neurologist without notification of hypokinetic-rigid symptoms in-between baseline and follow-up assessment. 2

Participants with CSVD and without parkinsonism 3 (n=503)

2 participants lost to follow-up Follow-up participants (n=501) 4

Renewed motor assessment (n=398) No renewed assessment of participants but clinical endpoints were obtained (n=103) • deceased (49) 5 Screened positive for parkinsonism (n=68) • unable to visit (54)

Analysis at Refused analysis; Participants with parkinsonism mentioned by Neurology OPD consensus meeting specialist; reviewed in consensus meeting (n=2) 6 (n=40) (n=28)

10 30 no 8 20 no parkinson parkinson parkinson parkinson 2 parkinsonism (ism) (ism) (ism) (ism) 7

7 7 1 VP VP VP 8 3 1 1 IPD IPD PSP

Total participants with incident parkinsonism 9 n = 20 (4%)

Figure 2. Flowchart of parkinsonism case finding during follow-up A Abbreviations: IPD: idiopathic Parkinson’s Disease; OPD: outpatient department; PSP: Progressive Supranuclear Palsy; CSVD: cerebral Small Vessel Disease; VP: vascular parkinsonism

81 CHAPTER 5

MRI scanning and processing Baseline MRI was performed on a single 1.5-Tesla Magnetom Sonata scanner (Siemens Medical Solutions, Erlangen, Germany), and included a 3D T1 magnetization-prepared rapid gradient echo, fluid-attenuated inversion recovery (FLAIR), gradient-echo T2*-weighted sequence and a DTI sequence. Details have been described in detail elsewhere.47 WMH were manually segmented on the FLAIR images, with a good inter-rater variability (intraclass correlation coefficient 0.99). The ratings of lacunes and microbleeds were revised according to the recently published Standards for Reporting Vascular Changes on Neuroimaging (STRIVE)5 by trained raters blinded to clinical information. The intra-rater and inter-rater variability in a random sample of 10% was good, with weighted kappa of 0.87 and 0.95, respectively, for presence of lacunes, and 0.85 and 0.86 for presence of microbleeds. Automated segmentation on T1 images was performed using Statistical Parametric Mapping (SPM5), to obtain grey matter (GM), WM and cerebrospinal fluid probability maps. These maps were binarized by applying a 0.5 threshold and summed to supply total volumes. All volumes were normalized to the total intracranial volume to adjust for head size.98 The DTI analysis has been described in detail elsewhere.47 For TBSS analysis, DTIFit within the FSL toolbox was used to generate fractional anisotropy (FA) and mean diffusivity (MD) images, which were imported into the TBSS pipeline.46 To create a FA skeleton, the mean FA image was thinned and subsequently this skeleton was thresholded at 0.3 to include major WM tracts. Of the 501 participants (2 were lost to follow-up), 4 were excluded from TBSS analysis because of imaging artefacts, 54 because of territorial infarcts, 2 because of missing values of microbleeds, and 5 because of parkinsonism other than VP, yielding a subgroup of 436 participants (9 with VP and 427 without parkinsonism).

Other measurements We used the Mini Mental State Examination (MMSE) score to indicate global cognitive status.

Statistical analysis Statistical analyses were performed using IBM (Armonk, NY) SPSS Statistics 20. The person- years at risk for each participant were defined as the time between baseline assessment and onset of parkinsonism, date of follow-up assessment, or death, depending on which event occurred first. Cumulative risk of (any) parkinsonism and separate for VP, being the largest group in our study, was estimated with a Kaplan-Meier analysis. Differences in baseline characteristics between participants with VP or IPD / PSP and without parkinsonism were tested by univariate analyses, using an independent samples t test, χ2 test, Fisher exact test, or Mann-Whitney U test, when appropriate (Table 1). Cox regression analysis was used to calculate hazard ratios (HR) with their corresponding 95% confidence intervals (CI) of baseline imaging characteristics for (any) parkinsonism and VP separately. Adjustments were made for baseline age, sex, UPDRS-m score, territorial infarcts, and for GM volume and or 4 CSVD characteristics (WMH volume, WM volume, number

82 BASELINE CEREBRAL SMALL VESSEL DISEASE AND INCIDENT PARKINSONISM.

of lacunes, and microbleeds). Verification of proportionality of hazards was performed by examining Schoenfeld residuals. Bonferroni corrections were used to correct for multiple comparisons; p≤0.00714 were considered significant. 1 To compare voxel-wise analyses of DTI measures between those with VP and without parkinsonism a 2-sample t test was performed, using a permutation-based statistical interference as part of FSL toolbox (randomise), with a standard number of permutation tests set at 5000. To identify significant associations, a threshold-free cluster enhancement with a 2 p<0.025, corrected for multiple comparisons, was used.

Table 1. Baseline characteristics of the total study population and of participants with VP, IPD/PSP and participants without parkinsonism 3 Total VP IPD/PSP No parkinsonism Baseline demographics n = 500 n = 14 n = 5 n = 481 Age, mean (SD), years 65.6 (8.8) 70.7 (6.3) 68.7 (8.1) 65.5 (8.8) Male sex, No. (%) 284 (56.8) 9 (64.3) 4 (80.0) 271 (56.3) 4 MMSE score, mean (SD) 28.1 (1.6) 27.4 (1.4) 27.0 (1.9) 28.2 (1.6) UPDRS-m total score, median (IQR) 0.0 (0.0-1.0)b 2.0 (0.0-6.0) 3.0 (2.0-4.0)b 0.0 (0.0-1.0) MRI measuresa n = 500 n = 14 n = 5 n = 481 WMH volume, median (IQR), mL 7.2 (3.6-18.4) 30.0 (16.6-56.9) 5.4 (3.7-22.1) 7.0 (3.4-17.7) 5 Lacunes presence, No. (%) 134 (26.8) 11 (78.6) 2 (40.0) 121 (25.2) Microbleeds presence, No. (%) 80 (16.0)c 8 (57.1) 0 (0.0) 72 (15.0)c White matter volume, mean (SD), mL 464.7 (51.9) 422.9 (63.4) 452.9 (33.8) 466.1 (51.2) Grey matter volume, mean (SD), mL 630.9 (53.9) 580.0 (48.2) 634.5 (76.7) 632.3 (53.2) Territorial infarcts presence, No. (%) 56 (11.2) 5 (35.7) 1 (20.0) 50 (10.4) 6 Baseline DTI measures, mean (SD) n = 497d n = 14 n = 5 n = 478d White matter global FA 0.33 (0.02) 0.31 (0.03) 0.34 (0.02) 0.33 (0.02) WMH global FA 0.34 (0.03) 0.31 (0.03) 0.35 (0.03) 0.34 (0.03) NAWM global FA 0.33 (0.02) 0.31 (0.03) 0.34 (0.02) 0.33 (0.02) 7 White matter global MD 0.89 (0.05) 0.95 (0.05) 0.87 (0.04) 0.89 (0.04) WMH global MD 1.00 (0.07) 1.09 (0.06) 0.99 (0.07) 1.00 (0.07) NAWM global MD 0.89 (0.04) 0.94 (0.04) 0.87 (0.03) 0.89 (0.04) Abbreviations: DTI: diffusion tensor imaging;FA : fractional anisotropy; IPD: idiopathic Parkinson disease; IQR: 8 interquartile range; MD: mean diffusivity (x10-3 mm2/s); MMSE: Mini-Mental State Examination; NAWM: normal- appearing white matter; PSP: progressive supranuclear palsy; SD: standard deviation; UPDRS-m: Unified Parkinson’s Disease Rating Scale motor score; VP: vascular parkinsonism; WMH: white matter hyperintensities. a Brain volumes are represented normalized to the total intracranial volume. b 1 participant was excluded because of a missing values on UPDRS-m at baseline. 9 c 4 participants were excluded because of missing values of microbleeds at baseline. d 3 participants were excluded because of baseline DTI artefacts.

83 CHAPTER 5

Results The total study population consisted of 501 participants; 2 were lost to follow-up. Mean follow-up duration was 5.2 years (SD 0.7). Parkinsonism developed in 20 participants (4.0%); 15 were diagnosed with VP, with all patients having predominantly lower body symptoms and a bilateral onset, 4 with IPD, and 1 with PSP. The cumulative 5-year risk of (any) parkinsonism was 3.5% (95% CI 1.9-5.2) and of VP 2.9% (95% CI 1.4-4.4). One participant with VP was excluded because of baseline T1/T2 artefacts. Table 1 shows the baseline characteristics of the total study population, and for participants with VP, IPD/ PSP, and without parkinsonism separately. The mean baseline age of the total population was 65.6 years (SD 8.8); 56.8% were men. Participants with VP, in comparison to participants without parkinsonism, were older (p=0.009), had a lower MMSE score (p=0.039), and a higher UPDRS-m score (p=0.001) at baseline. Furthermore, all baseline imaging characteristics shown in table 1 differed substantially p( <0.05) between those groups. For further analyses one participant with IPD was additionally excluded because of a missing baseline UPDRS-m score. There was a strong relation between WMH volume and the number of lacunes and the 5-year risk of (any) parkinsonism (Table 2). The 5-year risk of VP was increased for those with a high WMH volume (HR 2.0 per SD increase (mL); 95% CI 1.4-2.7), a high number of lacunes (HR 1.5 per number increase; 95% CI 1.2-1.9), presence of microbleeds (HR 5.7; 95% CI 1.9-16.8), and a low GM volume (HR 0.4 per SD increase (mL); 95% CI 0.2-0.8) (Table 3). A TBSS analysis showed differences in baseline DTI values between participants with VP and those without parkinsonism (Figure 3). Lower FA values were seen in WM tracts in the bilateral frontal and right parietal lobe – genu of corpus callosum, internal capsule, superior longitudinal fasciculus, forceps minor, inferior fronto-occipital fasciculus, cingulum bundle, superior and posterior (right) corona radiata and right posterior thalamic radiation – in participants with VP, even after adjustment for different CSVD characteristics. In addition, higher MD values were seen in VP patients in a similar pattern, although most voxels lost signal after adjustment for CSVD, except in the anterior corona radiata (data not shown).

84 BASELINE CEREBRAL SMALL VESSEL DISEASE AND INCIDENT PARKINSONISM. d d

: Unified value value 0.02 0.09 0.15 0.02 0.24 0.73 0.23 0.33 0.40 0.22 0.009 0.003 <0.001 p- p- 1 UPDRS-m b b

2 a a a a a a b a,b a,b a,b a,b a,b a,b : magnetic resonance imaging; resonance : magnetic MRI or CSVD characteristics or CSVD and CSVD characteristics and CSVD a a /s);

2 3 : cerebral small vessel disease; disease; small vessel : cerebral mm -4 Hazard Ratio (95% CI) Ratio Hazard matter grey for In addition adjusted volume 1.75 (1.31-2.35) 3.66 (1.19-11.32) 1.43 (1.13-1.80) 3.52 (1.36-9.09) 1.09 (0.99-1.19) 0.64 (0.34-1.18) 0.48 (0.27-0.87) (95% CI) Ratio Hazard matter grey for In addition adjusted volume 1.40 (0.80-2.44) 0.90 (0.49-1.66) 1.42 (0.81-2.49) 0.71 (0.36-1.42) 1.39 (0.4-3.01) 0.66 (0.35-1.27) CSVD

; d d d d d d

d 4 value -value p- <0.001 0.006 <0.001 0.005 0.14 0.27 0.004 p 0.28 0.04 0.30 0.05 0.001 0.17 : mean diffusivity (x10 : mean

MD 5

6 : fractional anisotropy; anisotropy; : fractional FA

7 : white matter hyperintensities. matter : white hazard ratios per standard deviation difference from the mean from difference deviation per standard ratios hazard MH

: diffusion tensor imaging; : diffusion 8 Hazard Ratio (95% CI) Ratio Hazard score baseline age, sex, UPDRS-m for Adjusted infarcts and territorial 1.74 (1.33-2.27) 4.74 (1.55-14.51) 1.52 (1.21-1.91) 3.87 (1.50-10.01) 1.07 (0.98-1.17) 0.73 (0.41-1.28) 0.42 (0.23-0.76) (95% CI) Ratio Hazard score baseline age, sex, UPDRS-m for Adjusted infarcts and territorial 0.76 (0.47-1.24) 0.54 (0.30-0.97) 0.77 (0.47-1.26) 1.70 (0.99-2.90) 2.49 (1.45-4.28) 1.48 (0.85-2.58) per SD: DTI c f (n = 499) e (n = 496)

e 9 : confidence interval; interval; : confidence CI

A normal-appearing white matter; white normal-appearing Significant after Bonferroni correction. Bonferroni after Significant Baseline MRI measures WMH volume, per SD, mL Lacunes, presence Lacunes, per number presence Microbleeds, per number Microbleeds, volume, per SD, mL matter White volume, per SD, mL Grey matter measures Baseline DTI per SD FA, global matter White per SD FA, WMH global per SD FA, NAWM global MD, per SD global matter White MD, per SD WMH global MD, per SD NAWM global In addition adjusted for CSVD characteristics: including white matter volume, WMH volume, number of lacunes, and microbleeds. number of lacunes, and microbleeds. WMH volume, volume, matter including white characteristics: CSVD for In addition adjusted In addition adjusted for grey matter volume. matter grey for In addition adjusted 4 Participants were excluded because of missing values of microbleeds at baseline. at of microbleeds of missing values because excluded were 4 Participants 1 Participant with IPD was excluded in addition because baseline UPDRS-m score was missing. was score UPDRS-m baseline in addition because excluded with IPD was 1 Participant 3 Participants were additionally excluded because of baseline DTI artefacts. DTI of baseline because additionally excluded were 3 Participants Table 2. Association between baseline MRI and DTI measures and the risk of (any) parkinsonism at follow-up follow-up parkinsonism at (any) the risk of and measures baseline MRI and DTI between Association 2. Table Abbreviations: Abbreviations: NAWM: W score; motor Scale Rating Disease Parkinson’s a b c d e f

85 CHAPTER 5 d d d d

: Unified 0.03 0.04 0.07 0.73 0.69 0.68 0.65 0.21 0.51 -value -value 0.002 0.002 0.007 <0.001 p p UPDRS-m b b a a a a a a b a,b a,b a,b a,b a,b a,b : magnetic resonance imaging; resonance : magnetic MRI CSVD characteristics CSVD or and CSVD characteristics and CSVD /s); a a 2 : cerebral small vessel disease; disease; small vessel : cerebral mm -4 Hazard Ratio (95% CI) Ratio Hazard matter grey for In addition adjusted volume 1.99 (1.44-2.73) 4.68 (1.18-18.56) 1.49 (1.16-1.92) 5.68 (1.92-16.84) 1.10 (1.01-1.21) 0.51 (0.24-1.07) 0.39 (0.19-0.77) (95% CI) Ratio Hazard matter grey for In addition adjusted volume 1.12 (0.59-2.11) 0.87 (0.42-1.78) 1.14 (0.60-2.17) 0.83 (0.39-1.80) 1.81 (0.72-4.56) 0.79 (0.38-1.62) CSVD

; d d d d d d d d -value -value p <0.001 0.007 <0.001 0.001 0.08 0.18 0.001 p 0.02 0.007 0.02 0.002 <0.001 0.01 : mean diffusivity (x10 : mean MD : fractional anisotropy; anisotropy; : fractional FA : white matter hyperintensities. matter : white hazard ratios per standard deviation difference from the mean from difference deviation per standard ratios hazard WMH Hazard Ratio (95% CI) Ratio Hazard score baseline age, sex, UPDRS-m for Adjusted infarcts and territorial 1.92 (1.45-2.55) 6.60 (1.68-25.89) 1.61 (1.26-2.05) 6.52 (2.21-19.22) 1.08 (0.99-1.18) 0.64 (0.33-1.23) 0.32 (0.16-0.64) (95% CI) Ratio Hazard score baseline age, sex, UPDRS-m for Adjusted infarcts and territorial 0.52 (0.30-0.89) 0.39 (0.19-0.77) 0.53 (0.31-0.91) 2.49 (1.40-4.43) 3.77 (2.02-7.04) 2.20 (1.19-4.06) : diffusion tensor imaging; : diffusion per SD: DTI c f (n = 495) e (n = 492) e : confidence interval; interval; : confidence CI normal-appearing white matter; white normal-appearing Significant after Bonferroni correction. Bonferroni after Significant Baseline MRI measures WMH volume, per SD, mL Lacunes, presence Lacunes, per number presence Microbleeds, per number Microbleeds, volume, per SD, mL matter White volume, per SD, mL Grey matter measures Baseline DTI per SD FA, global matter White per SD FA, WMH global per SD FA, NAWM global MD, per SD global matter White MD, per SD WMH global MD, per SD NAWM global In addition adjusted for CSVD characteristics: including white matter volume, WMH volume, number of lacunes, and microbleeds. number of lacunes, and microbleeds. WMH volume, volume, matter including white characteristics: CSVD for In addition adjusted In addition adjusted for grey matter volume. matter grey for In addition adjusted 4 Participants were excluded because of missing values of microbleeds at baseline. at of microbleeds of missing values because excluded were 4 Participants 5 Participants were excluded because of a diagnosis of parkinsonism other than vascular parkinsonism. than vascular other of a diagnosis parkinsonism because excluded were 5 Participants artefacts. DTI of baseline in addition because excluded were 3 Participants Table 3. Association between baseline MRI and DTI measures and the risk of vascular parkinsonism at follow-up follow-up parkinsonism at vascular the risk of and measures baseline MRI and DTI between Association 3. Table Abbreviations: NAWM: score; motor Scale Rating Disease Parkinson’s a b c d e f

86 BASELINE CEREBRAL SMALL VESSEL DISEASE AND INCIDENT PARKINSONISM.

Discussion This is a unique prospective study investigating the relation between CSVD at baseline and the risk of incident parkinsonism. We showed that a high WMH volume and a high number of 1 lacunes were associated with an increased 5-year risk of (any) parkinsonism. For VP, this risk was also increased by the presence of microbleeds and a low GM volume. Furthermore, we observed lower FA values especially in bifrontal WM tracts involved in movement control in participants with VP compared to participants without parkinsonism, independent of CSVD. 2 Major strengths of our study are its longitudinal and single-centre design, which allowed us to use identical motor and cognitive assessments during baseline and follow-up. Furthermore, the large sample size and high follow-up rate of 99.6% are main advantages. Moreover, all imaging data were analyzed by raters blinded to clinical information with a good intra-rater 3 and inter-rater variability. Finally, we were able to make appropriate adjustments, reducing the risk of confounding. We intentionally did not adjust for vascular risk factors as we considered them part of the causal chain between CSVD and parkinsonism. Several methodological issues need to be addressed. First, because of the small number 4 of patients with parkinsonism in our study, our results should be interpreted with caution. Second, we were not able to diagnose parkinsonism in the same way for all patients. For participants who were not able to participate in person, we had to rely on information from medical files and we could have missed the diagnosis parkinsonism in some, since 5 parkinsonism is frequently accepted as part of normal aging. Some participants who participated in-person and were considered screen-positive for the presence of parkinsonian signs, refused additional evaluation by a neurologist specialized in movement disorders (28 of 68 participants). However, all 28 participants were examined in the follow-up assessment 6 by 2 skilled neurologists in training with extensive experience in diagnosing parkinsonism, after which a consensus diagnosis was made by an expert panel. The similarity of the diagnostic approaches is well-illustrated by a virtual identical proportion of the patients with parkinsonism identified by the 2 approaches. Third, misclassification could have occurred 7 as the accuracy of clinical diagnosis of the different aetiologies underlying parkinsonism, compared to neuropathologic diagnosis, is relatively low.130 We therefore initially classified any parkinsonism; thereafter, neuroimaging was used to allow for the diagnosis of VP. Fourth, 8 as imaging information is needed to classify the different aetiologies of parkinsonism, especially concerning VP, circular reasoning might have occurred in our analyses with VP, although we used baseline MRI and DTI measures, when all participants were free of parkinsonism. 9 Even though this is a hospital-based cohort study, our results have a high external validity for an elderly population with CSVD who visit a general neurology department, as we included all consecutive patients with CSVD on neuroimaging (CT or MRI-scan) performed because of major referral reasons (e.g., TIA, stroke, cognitive complaints) and there were no restrictions A for admission to our hospital.

87 CHAPTER 5

B p<0.01

p<0.025

x = --11 R yy= = --1717 L R zz= = 19 LL

Figure 3. Differences in fractional anisotropy values between participants with vascular parkinsonism and without parkinsonism Voxel-wise analysis of the differences in fractional anisotropy (FA) values between participants with vascular parkinsonism (n=9) and without parkinsonism (n=427). Adjusted for age, sex, baseline motor part of the Unified Parkinson’s Disease Rate Scale score, and normalized total brain volume (A) and for cerebral small vessel characteristics (white matter volume, white matter hyperintensity volume and number of lacunes and microbleeds) (B), performed with a 2 sample t test, thresholded at p <0.025 and corrected for multiple comparisons. These images are superimposed onto the spatially normalized Montreal Neurological Institute (MNI) stereotactic space FA map. R indicates right side, L indicates left side. The x, y and z coordinates represent the MNI coordinates of each slide.

88 BASELINE CEREBRAL SMALL VESSEL DISEASE AND INCIDENT PARKINSONISM.

So far, the exact role of CSVD in parkinsonism is unknown. Some studies have suggested that WMH are more common in patients with parkinsonism,131, 132 whereas others failed 133, 134 to demonstrate that. Furthermore, in autopsy studies, only a small subset (<10%) of 1 parkinsonism was attributed to vascular lesions, because of the absence of other pathologic findings (Lewy bodies or tau inclusions) compatible with a known parkinsonian syndrome.130 Our findings favour a role for CSVD in the development of parkinsonism, as we showed that participants with a high degree of CSVD had an increased risk of incident parkinsonism. 2 Furthermore, we found a relative high incidence of parkinsonism compared to population- based studies; in the Rotterdam Study, 2% of the 6566 participants (≥55 years) developed (any) parkinsonism after a mean follow-up duration of 5.8 years,129 versus 4% in our study, which may also indicate that CSVD contributes to the aetiology of parkinsonism. 3 DTI has gained increased interest in the diagnostic process of parkinsonism in recent years,121 as it has been suggested that this technique could be of help in differentiating among the different subtypes of parkinsonism.135, 136 We therefore performed a TBSS analysis in patients with a clinical diagnosis of VP, because this was the largest group of patients in our study and 4 a recent study showed that DTI can differentiate between VP and parkinsonian syndromes of degenerative origin.136 We found lower FA values in bilateral WM tracts involved in movement control in VP compared to participants without parkinsonism, even after adjustment for CSVD characteristics. This result is in line with a recent cross-sectional DTI study in patients with VP 5 and healthy controls; however, no adjustments were made for CSVD.122 Using DTI measures of scans at a point in time when all participants were free of parkinsonism is unique in our study. Our results may suggest that diffusion changes can be an early marker of VP. However, a note of caution is due here owing to the small number of patients with VP in our study. 6 We can hypothesize about the role of CSVD in parkinsonism; it may be that CSVD disrupts the structural integrity of WM tracts, including disruption of the thalamocortical fibres, thereby reducing the influence of the basal ganglia on motor, premotor and supplementary motor cortices.137 Disconnection of the basal ganglia-thalamocortical circuit possibly leads to (sub) 7 cortical atrophy, ultimately resulting in parkinsonism. Furthermore, CSVD could possibly lower the threshold for developing parkinsonism, by lowering the threshold for Lewy body pathology to become symptomatic, for example. In addition, marked loss of striatal 8 dopaminergic innervations that occurs during aging might contribute as well.138 Future studies are needed to further investigate the contribution of CSVD to incident parkinsonism, ideally taking into account the changes in these imaging markers over time. 9

Part IV Long-term mortality in cerebral small vessel disease

6. Factors associated with 8-year mortality in cerebral small vessel disease

Published as: H.M. van der Holst, I.W.M. van Uden, A.M. Tuladhar, K.F. de Laat, A.G. van Norden, D.G. Norris, E.J. van Dijk, L.C. Rutten-Jacobs, F-E de Leeuw.

Factors associated with 8-year mortality in older patients with cerebral small vessel disease: The Radboud University Nijmegen Diffusion tensor and Magnetic resonance Cohort (RUN DMC) study. JAMA Neurology, 2016 Apr; 73(4):402-9. CHAPTER 6

Abstract

Importance: Gait and cognition have been related to mortality in population-based studies. This is possibly mediated by cerebral small vessel disease (CSVD), which has been associated with mortality as well. It is unknown which parameters can predict mortality in individuals with CSVD. Identification of high-risk patients may provide insight into factors that reflect their vital health status. Objective: To assess mortality in patients with CSVD and to identify potential clinical and/or imaging predictors of mortality. Design, setting, and participants: A prospective, single-centre cohort study was conducted. The present investigation is embedded in the Radboud University Nijmegen Diffusion Tensor and Magnetic Resonance Cohort (RUN DMC) study. Between January 17, 2006, and February 27, 2007, all participants underwent a cognitive and motor assessment and cerebral magnetic resonance imaging (MRI) including a diffusion tensor imaging sequence to assess microstructural integrity of the white matter. Participants were followed until their death or November 24, 2014. Participants included 503 older adults with CSVD noted on brain imaging. Data analysis was performed from November 26, 2014, to February 2, 2015. Main outcomes and measures: Eight-year all-cause mortality. Results: Of 503 participants (mean [SD] age, 65.7 [8.8] years; range, 50-85 years; 284 [56.5%] were male), 80 individuals (15.9%) died during a mean (SD) follow-up of 7.8 (1.5) years. In the final analysis, 494 (98.2%) were included, of whom 78 (15.8%) died. Gait speed, cognitive index, conventional MRI markers of CSVD (white matter hyperintensity volume, brain volume, and lacunes), and diffusion measures of the white matter were associated with an 8-year risk of mortality independent of age, sex, and vascular risk factors. The prediction of mortality was determined using Cox proportional hazards models with backward stepwise selection and including age, sex, vascular risk factors, gait speed, cognitive index, MRI, and diffusion measures. Results are reported as hazard ratios (HRs) (95%CI). Older age (1.05 per 1-year increase [1.01-1.08]), lower gait speed (1.15 per 0.1- m/s slower gait [1.06-1.24]), lower grey matter volume (0.72 per 1-SD increase [0.55-0.95]), and greater global mean diffusivity of the white matter (1.51 per 1-SD increase [1.19-1.92]) were identified as the main factors associated with mortality. Cognitive index and other conventional CSVD markers were not retained in the prediction model. Conclusions and relevance: Gait, cognition, and imaging markers of CSVD are associated with 8-year risk of mortality. In the prediction of mortality, an older age, lower gait speed, lower grey matter volume, and greater global mean diffusivity of white matter at baseline best predicted mortality in our population. Further research is needed to investigate the reproducibility of this prediction model and to elucidate the association between the factors identified and mortality.

94 FACTORS ASSOCIATED WITH 8-YEAR MORTALITY IN CEREBRAL SMALL VESSEL DISEASE.

Introduction Cerebral small vessel disease (CSVD) is prevalent on brain imaging of older adults.9 It consists of white matter hyperintensities (WMH) and lacunes of presumed vascular origin, microbleeds 1 and subcortical and cortical atrophy on conventional magnetic resonance imaging (MRI).5 The radiological spectrum of CSVD extends beyond lesions visible on conventional MRI, including impaired white matter (WM) microstructural integrity which can be assessed by diffusion tensor imaging (DTI).91 The clinical presentation and long-term prognosis of CSVD 2 are both highly variable, including cognitive and motor impairment and mood disturbances, which could lead to functional decline139 and even death.22, 23 It is unknown which patients with CSVD are at the highest risk for these adverse outcomes, including mortality. Several population-based studies among community-dwelling older adults have shown that gait 3 speed25 and cognition24, 140 are important clinical characteristics associated with mortality; nevertheless, to our knowledge, whether this prediction was independent of the presence of CSVD has not been investigated. We could hypothesize that this association is driven by CSVD since CSVD has been previously associated with mortality, particularly, WMH and lacunes.23, 141 4 However, to our knowledge, it has never been investigated whether imaging characteristics of CSVD have added value in the prediction of mortality regarding clinical factors such as gait and cognition. In this observational study, we prospectively investigated the cumulative mortality in 5 a population of older adults with CSVD after 8 years of follow-up. Our main objective was to identify baseline risk factors of all-cause mortality, including clinical (gait speed and cognition) and imaging (MRI and DTI measures) factors. We were especially interested in which of these parameters were most predictive for all-cause mortality. This information may 6 provide insight into factors that reflect the vital health status of older adults with CSVD.

95 CHAPTER 6

Methods Study population This present study is embedded in the Radboud University Nijmegen Diffusion tensor and Magnetic resonance Cohort study (RUN DMC study), a prospective cohort study that investigates the risk factors and clinical consequences of functional and structural brain changes as assessed by MRI. A total of 503 older adults with CSVD aged 50 to 85 years were included. The recruitment, study rationale and protocol of the RUN DMC study have been reported.47 Because the onset of CSVD is often insidious and clinically heterogeneous with acute symptoms (transient ischemic attacks or lacunar syndromes), or subacute symptoms (cognitive, motor and mood disturbances),123 an CSVD diagnosis was made on the basis of brain imaging and included the presence of WMH and/or lacunes of presumed vascular origin.48 Major referral reasons of the participants to our department included those corresponding to symptoms of CSVD (e.g., transient ischemic attacks or minor stroke and cognitive disturbances). After inclusion, all participants were subsequently asked about acute or subacute symptoms of CSVD. Main exclusion criteria were baseline parkinsonism, dementia, life expectancy of less than 6 months, non-CSVD related WM lesions (e.g., multiple sclerosis), and MRI contra-indications. Subsequently, the above-mentioned acute or subacute clinical symptoms of CSVD were assessed by standardized structured assessments. Patients who were eligible because of a lacunar syndrome were included more than 6 months after the event to avoid acute effects on the outcomes. Baseline assessment included a cognitive and motor examination and a cerebral MRI. The present study was conducted from January 17, 2006, to February 27, 2007. Participants were followed until their death or until November 24, 2014. The Medical Review Ethics Committee region Arnhem-Nijmegen approved the study. All participants signed an informed consent form; there was no financial compensation.

Mortality All-cause mortality was the primary outcome of this study. Information on vital status was first retrieved from the Dutch Municipal Personal Records database (https://www.government.nl /topics/identification-documents/contents/the-municipal-personal-records-database). Information on the cause of death was obtained from the general practitioner or treating physician and medical records. Subsequently, the cause of death was classified according to the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10)142 by 1 rater (H.M. v.d. H.) and grouped as ischemic stroke, intracranial haemorrhage, cardiac cause, other vascular causes (when the cause of death was presumably vascular, but did not meet the criteria for fatal stroke or cardiac cause), malignant neoplasms, infections and miscellaneous.

96 FACTORS ASSOCIATED WITH 8-YEAR MORTALITY IN CEREBRAL SMALL VESSEL DISEASE.

MRI scanning and processing All participants underwent a cerebral MRI, including a 3-dimensional T1 magnetization prepared rapid acquisition gradient-echo (MPRAGE), fluid-attenuated inversion recovery 1 (FLAIR), gradient-echo T2*-weighted sequence and a DTI sequence on a 1.5-Tesla scanner at baseline. Imaging details have been described elsewhere.47 WMH were manually segmented on the FLAIR images and total WMH volume was calculated by summing all segmented areas multiplied by section thickness, with a good inter-rater 2 variability (intra-class correlation coefficient, 0.99). The ratings of lacunes and microbleeds (including subcortical and lobar ones) were revised according to the recently published Standards for Reporting Vascular changes on neuroimaging5 by trained raters (including H.M. v.d. H. and I.W.M. v. U.) blinded to clinical information (intra-rater and inter-rater reliabilities: 3 weighted ĸ values, 0.87 and 0.95, respectively, for presence of lacunes and 0.85 and 0.86, respectively, for presence of microbleeds, calculated in a random sample of 10% of the scans). Automated segmentation on T1 images was performed using Statistical Parametric Mapping, 4 version 5 (SPM5; http://www.fil.ion.ucl.ac.uk/spm/software/), to obtain grey matter (GM) and WM and cerebrospinal fluid probability maps. The volumes were calculated by summing all the voxel volumes belonging to the tissue class. All volumes, including WMH were normalized to the total intracranial volume (sum of GM, WM, and cerebrospinal fluid volumes)98 to adjust 5 for head size. GM volume was composed of the volume of the neocortex, basal ganglia and thalamus. For DTI analysis, the diffusion-weighted images of each participant were realigned on the mean of the unweighted image using mutual information-based coregistration routines from 6 SPM5. The diffusion tensor32 and its eigenvalues were estimated using linear regression and spurious negative eigenvalues were set to zero, after which the tensor derivates of fractional anisotropy and mean diffusivity (MD) were calculated.99 The mean unweighted image was used to compute the coregistration variables to the anatomical T1 reference image, which 7 were then applied to all diffusion-weighted images and results. All images were visually checked for motion artefacts and coregistration errors. The volume-averaged fractional anisotropy and MD were calculated in the total WM. 8 Cardiovascular risk factors Information on the presence of cardiovascular risk factors was investigated with structured questionnaires. The use of medication for treatment of any vascular risk factor was verified 9 by a medication list from the pharmacy provided by the participant. Information on smoking behaviour was dichotomized into ever (current and former) and never smoking. Diabetes mellitus was considered to be present if the participant was receiving oral glucose-lowering drugs or insulin. Hypertension was defined as the use of A blood pressure-lowering medication and/or a current systolic blood pressure of 140 mmHg or higher or diastolic blood pressure of 90 mmHg or higher, assessed during baseline

97 CHAPTER 6 examination, with the mean determined after 3 measurements with the patient in a supine position after 5 minutes of rest.143

Cardiovascular diseases and malignant neoplasms To identify cardiovascular disease or malignant neoplasms in the medical history, structured questionnaires were used and this information was subsequently retrieved by their treating physician or from medical files and was verified thereafter. A history of cardiovascular diseases was defined as the presence of an ischemic stroke, intracerebral haemorrhage, transient ischemic attack, myocardial infarction, percutaneous coronary intervention, coronary bypass surgery, or peripheral arterial disease. A history of cancer was defined as the presence of any malignant neoplasm (mentioned in the ICD-10).142

Measurement of gait Gait speed was assessed by using a 5.6-m electronic portable walkway (GAITRite; MAP/ CIR Inc). This walkway system has an excellent test-retest reliability and validity.93, 94 Each participant was instructed to walk over the walkway at his or her usual speed. Participants started 2 m before the walkway and stopped 2 m behind it to measure steady-state walking. The mean gait speed (meters per second) of two walking episodes of each participant was used for analysis.

Cognitive assessment Global cognitive function was evaluated by the Mini Mental State Examination.50 A Cognitive Index was constructed to obtain a more robust outcome measure for global cognition. The cognitive index was calculated as the mean of the z-scores of the Speed Accuracy Tradeoff score of the 1-letter subtask of the Paper-Pencil Memory Scanning Task,54 the mean of the Speed Accuracy Tradeoff score of the reading task of the Stroop test,52 the mean of the Symbol-Digit Substitution task,55 and the mean of the added score on the 3 learning trials of the Rey Auditory Verbal learning test and the mean of the delayed recall of this.51, 100

Statistical analysis Statistical analyses were performed with IBM SPSS Statistics 20 for Windows (SPSS Inc) and R, version 2.15 (http://www.R-project.org) software packages. Cumulative mortality was estimated using Kaplan-Meier analysis and stratified for different imaging markers. Differences between the lowest and highest quartiles and presence versus absence were estimated with the log-rank test. Data analysis was performed from November 26, 2014, to February 2, 2015. Differences in baseline characteristics between survivors and those who died were tested by using an independent samples t test, χ2 test, or Mann-Whitney test when appropriate. Cox regression analysis was used to calculate hazard ratios (HR) and 95% confidence intervals (CI) of gait speed, cognitive index and different imaging measures for mortality after

98 FACTORS ASSOCIATED WITH 8-YEAR MORTALITY IN CEREBRAL SMALL VESSEL DISEASE. adjustment for age, sex and vascular risk factors. For the prediction models, we used the variables with a significance level of p≤0.10 after adjustment for age, sex and vascular risk factors. Because of the high correlation between DTI variables, only MD of the WM was included 1 in the model. Three Cox proportional hazards models were constructed to predict all-cause mortality. In the first model, age, sex and vascular risk factors were entered simultaneously into the Cox model. For the next models these factors were fixed. Subsequently, gait speed and cognitive index (model 2) and imaging parameters (MRI and DTI measures, model 3) 2 were entered by a backward, stepwise selection procedure until these nonfixed variables had a significance level of p≤0.10. A fourth model was constructed by using a backward, stepwise selection procedure for all covariates (age, sex, vascular risk factors, cognition, gait and MRI and DTI measures); none of these were fixed. Models 1 to 3 were also constructed 3 for vascular mortality; using proportional hazard model by means of Fine and Grey,144 causes of death other than vascular factors were considered a competing risk. Schoenfeld residuals were investigated to verify proportionality of hazards. There were no indications that the proportional hazard assumption was violated. 4 The C-statistic was used to assess the discriminatory performance of the different prediction models. An increase of C-statistic values by 0.025 or more was considered a significant improvement of accuracy.25 Furthermore, Akaike information criterion (AIC) was used to investigate the goodness of fit of the models.145 The most appropriate model is the one 5 with the lowest AIC value.146 We considered a decrease of 10 or more values as significantly improved goodness of fit.

99 CHAPTER 6

Table 1. Baseline characteristics by vital status of the study population Vital status at follow-up Alive Dead p-value Characteristics (n = 416) (n = 78) Baseline demographics Age, mean (SD), years 64.5 (8.4) 72.1 (8.4) <0.001a Male sex, No. (%) 236 (56.7) 43 (55.1) 0.79b Vascular risk factors at baseline, No. (%) Smoking 286 (68.8) 60 (76.9) 0.15b Diabetes mellitus 44 (10.6) 19 (24.4) 0.001b Hypertension 299 (71.9) 64 (82.1) 0.06b Co-morbidity at baseline, No. (%) Cardiovascular morbidity 195 (46.9) 48 (61.5) 0.02b Malignant neoplasm 42 (10.1) 10 (12.8) 0.47b Baseline clinical scores Gait velocity, mean (SD), m/s 1.32 (0.26) 1.09 (0.28) <0.001a MMSE score, mean (SD) 28.3 (1.6) 27.5 (1.8) <0.001c Cognitive index, mean (SD) 0.08 (0.76) -0.52 (0.68) <0.001a Baseline MRI characteristicsd WMH volume, median (IQR), mL 6.0 (3.2-15.2) 17.7 (9.8-35.1) <0.001c WM volume, mean (SD), mL 470.5 (50.7) 433.6 (41.8) <0.001a GM volume, mean (SD), mL 636.9 (52.9) 597.6 (47.1) <0.001a Lacunes present, No. (%) 92 (22.1) 41 (52.6) <0.001b Microbleeds, present, No. (%)e 62 (14.9) 19 (24.4) 0.03b Territorial infarcts present, No. (%) 42 (10.1) 13 (16.7) 0.09 b Baseline DTI parameters WM global FA, mean (SD) 0.33 (0.02) 0.32 (0.02) <0.001a WM global MD, mean (SD), x 10-4 mm2/s 8.8 (0.4) 9.3 (0.4) <0.001a

Abbreviations: DTI, diffusion tensor imaging; FA, fractional anisotropy; GM, grey matter; IQR, interquartile range; MD, mean diffusivity; MMSE, Mini-mental State Examination; MRI, magnetic resonance imaging; SD: standard deviation; WM, white matter; WMH, WM hyperintensities. a Independent samples t-test. b Chi-square test. c Mann-Whitney Test. d Brain volumes are represented as normalized to the total intracranial volume. e Two participants were excluded because of missing values of microbleeds at baseline (1 in each group)

100 FACTORS ASSOCIATED WITH 8-YEAR MORTALITY IN CEREBRAL SMALL VESSEL DISEASE.

Results The study population consisted of 503 participants; 493 individuals (98.0%) were white, with a mean (SD) age at baseline of 65.7 (8.8) years and a mean follow-up duration of 7.8 (1.5) 1 years. Eighty participants (15.9%) died during the follow-up period. The 8-year cumulative all-cause mortality was 14.5% (95% CI, 11.3-17.6). Nine participants were excluded because of imaging artefacts (4) and missing values on gait or cognitive tests (5); 2 of these 9 individuals had died. Table 1 reports the baseline characteristics of the study population (494 [98.2% 2 of the original population]). The cause of death was vascular related in 26 (33.3%) of the participants (Table 2).

Table 2. Cause of death according to ICD-10 classification 3 Cause of death Participants, No. (%) Vascular 26 (33.3) Ischemic stroke 5 (6.4) Intracerebral haemorrhage 3 (3.8) 4 Cardiac cause 12 (15.4) Other vasculara 6 (7.7) Malignancies 20 (25.6) Infections 12 (15.4) 5 Miscellaneous 15 (19.2) Unknown 5 (6.4)

Abbreviation: ICD-10, International Statistical Classification of Diseases and Related Health Problems, Tenth Revision 6 a Deaths that were presumably vascular but did not meet the criteria for fatal stroke or cardiac cause

The Figure shows the cumulative mortality stratified for different imaging characteristics. All- 7 cause 8-year mortality was highest in participants with lacunes and microbleeds, with the highest WMH volume and MD of the WM and the lowest WM and GM volumes. Gait speed, cognitive index, conventional MRI markers of CSVD (except for microbleeds) and 8 diffusion measures of the WM were identified as potential risk factors of all-cause mortality (Table 3). The presence of territorial infarcts was not associated with mortality.

101 CHAPTER 6

Lacunes White matter hyperintensity volume 40 40 presence lowest quartile second quartile 30 absence 30 third quartile

highest quartile 20 20

p<0.001 10 p<0.001 cumulativemortality (%) 10 Cumulativemortality (%)

0 0 0 2 4 6 8 10 0 2 4 6 8 10 Time (years) Time (years)

no. at risk 361 360 354 338 277 no. at risk 123 123 123 121 108 133 127 117 107 79 124 124 123 121 97

124 124 119 108 74 123 116 106 95 77

White matter volume Microbleeds 40 40 lowest quartile

presence second quartile 30 30 absence third quartile highest quartile 20 20

Cumulativemortality (%) 10 p<0.001 10 p=0.03 Cumulativemortality (%)

0 0 0 2 4 6 8 10 0 2 4 6 8 10 Time (years) Time (years)

no. at risk 123 118 110 100 75 no. at risk 411 407 397 377 307 124 124 119 106 84 81 78 72 67 49 124 124 122 119 92 123 121 120 120 105

Gray matter volume Mean diffusivity of the global white matter

40 40

lowest quartile lowest quartile

second quartile second quartile 30 30 third quartile third quartile

highest quartile highest quartile 20 20 cumulativemortality (%) cumulativemortality (%) 10 p<0.001 10 p<0.001

0 0 0 2 4 6 8 10 0 2 4 6 8 10 Time (years) Time (years)

no. at risk 123 122 114 99 75 no. at risk 123 123 123 121 97 124 118 114 110 88 124 124 124 121 97 124 124 121 115 91 124 122 117 113 93 123 123 122 121 102 123 118 107 90 69

Figure. Cumulative mortality stratified by imaging characteristics Cumulative mortality was estimated using Kaplan-Meier analysis; this was stratified for different magnetic resonance imaging markers. The differences between the lowest and highest quartiles and presence versus absence were estimated with log-rank test.

102 FACTORS ASSOCIATED WITH 8-YEAR MORTALITY IN CEREBRAL SMALL VESSEL DISEASE.

Table 3. Association between baseline factors and all-cause mortality

Characteristic HR (95% CI)a p-valueb HR (95% CI)c p-valueb Clinical scores 1 Gait speed, per 0.1m/s slower gait 1.19 (1.10-1.28) <0.001 1.17 (1.08-1.26) <0.001 MMSE 0.89 (0.79-1.01) 0.07 0.91 (0.81-1.03) 0.15 Cognitive index 0.54 (0.38-0.77) 0.001 0.59 (0.41-0.84) 0.004 MRI measuresd 2 WMH volume, per 1-SD increase, mLe 1.65 (1.28-2.15) <0.001 1.62 (1.24-2.11) <0.001 WM volume, per 1-SD increase, mL 0.71 (0.54-0.92) 0.009 0.74 (0.57-0.97) 0.03 GM volume, per 1-SD increase, mL 0.61 (0.47-0.81) 0.001 0.65 (0.49-0.86), 0.003 No. of lacunes 1.28 (1.09-1.49) 0.002 1.23 (1.05-1.44) 0.01 3 No. of microbleedsf 1.02 (0.96-1.09) 0.52 1.01 (0.95-1.09) 0.69 Territorial infarcts, presence 1.39 (0.77-2.54) 0.28 1.25 (0.69-2.29), 0.46 DTI parameters WM global FA, per 1-SD increase 0.70 (0.57-0.87) 0.001 0.70 (0.56-0.87) 0.002 WM global MD, per 1-SD increase 1.70 (1.35-2.14) <0.001 1.68 (1.32-2.13) <0.001 4

Abbreviations: CI: confidence interval; DTI: diffusion tensor imaging;FA: fractional anisotropy; GM: grey matter; HR: hazard ratio; MD: mean diffusivity (*10-4 mm2/s); MMSE: Mini-mental State Examination; MRI: magnetic resonance imaging; SD: standard deviation; WM: white matter; WMH: WM Hyperintensity. 5 a Adjusted for age and sex. b Cox regression analysis. c Adjusted additionally for vascular risk factors (smoking, diabetes mellitus and hypertension). d Brain volumes are represented normalized to the total intracranial volume. e Log transformed. 6 f Two participants were excluded because of missing values of microbleeds at baseline.

103 CHAPTER 6

In the prediction of all-cause mortality (Table 4), model 2 shows that gait speed and cognitive index added substantially to model 1. In model 3, imaging parameters were also included; GM volume and global MD of the WM were retained in the model together with gait speed. Cognitive index and other imaging markers were not retained in the prediction model. This last model predicted the best mortality in our population. When we applied backward stepwise selection for all covariates mentioned in model 3, the following covariates were retained (reported as HR [95% CI]): age (1.05 per 1-year increase [1.01-1.08]; p=0.01), gait speed (1.15 per 0.1-m/s slower gait [1.06-1.24]; p=0.001), GM volume (0.72 per 1-SD increase [0.55-0.95]; p=0.02), and global MD of the WM (1.51 per 1-SD increase [1.19-1.92]; p=0.001). This model was as accurate as model 3 (C-statistic, 0.79 [95% CI, 0.72-0.85] and AIC, 854.8). When cardiovascular morbidity and malignant neoplasms were additionally added as fixed covariates, the above-mentioned prediction models were not substantially altered. Hazard ratios of the same magnitude were found for vascular mortality compared with all- cause mortality (model 3 with age, sex and vascular risk factors as fixed factors; reported as HR [95% CI])): gait speed (1.11 per 0.1-m/s slower gait [0.97-1.28]; p=0.13), GM volume (0.72 per 1-SD increase [0.45-1.13]; p=0.15), and the MD of the WM (1.85 per 1-SD increase [1.29- 2.65]; p<0.001), with other causes of death considered a competing risk. However, because of the small number of patients (n=26), significance was lost for gait speed and GM volume and only the MD of the WM was retained in the model (Table 5).

104 FACTORS ASSOCIATED WITH 8-YEAR MORTALITY IN CEREBRAL SMALL VESSEL DISEASE. magnetic magnetic

MRI: 1 value p- 0.003 0.25 0.11 0.10 0.27 0.002 0.04 0.001

2 Mean Diffusivity; Mean MD: WM hyperintensity. WM hyperintensity. Model 3 1.06 (1.02-1.09) 0.73 (0.43-1.25) 1.65 (0.89-3.04) 1.60 (0.91-2.80) 0.71 (0.38-1.32) 1.13 (1.04-1.22) NR NR NR 0.75 (0.56-0.99) NR 1.53 (1.19-1.96) 0.80 (0.73-0.86) 856.4 3 WMH: hazard ratio; ratio; hazard HR: value p- <0.001 0.31 0.09 0.08 0.68 0.006 0.09

white matter; matter; white 4 grey matter; matter; grey WM: GM:

Model 2 1.08 (1.05-1.12) 0.76 (0.45-1.29) 1.69 (0.93-3.09) 1.64 (0.94-2.88) 0.88 (0.48-1.61) 1.13 (1.04-1.23) 0.72 (0.49-1.05) NI NI NI NI NI 0.77 (0.71-0.84) 868.8 5 : standard deviation; deviation; : standard SD value p- <0.001 0.07 0.046 0.006 0.88 diffusion tensor imaging; diffusion 6 DTI:

7 not retained in the model; retained not HR (95% CI) Model 1 1.11 (1.08-1.14) 0.63 (0.38-1.04) 1.83 (1.01-3.32) 2.14 (1.24-3.68) 0.96 (0.53-1.74) NI NI NI NI NI NI 0.74 (0.68-0.81) 880.6 NI NR: : confidence interval; interval; : confidence CI /s 2 8 a mm -4

9 I: not included in the model; I: not N Akaike information criterion; criterion; information Akaike

A Log transformed Log Characteristic (n = 494) Characteristic Demographics per 1-y increase Age, Male sex risk factors Vascular Smoking mellitus Diabetes Hypertension scores Clinical Gait speed, per 0.1m/s slower gait Cognitive index measures MRI and DTI mL increase, WMH volume, per 1-SD mL increase, WM volume, per 1-SD mL increase, GM volume, per 1-SD increase Lacunes, per No. x10 increase, MD of WM, per 1-SD (95% CI) C-statistic AIC Table 4. Baseline indicator variables retained in Cox regression models for prediction of all-cause mortality (n = 78) = 78) (n mortality all-cause of prediction for models regression in Cox retained variables 4. Baseline indicator Table AIC: Abbreviations: imaging; resonance a

105 CHAPTER 6 , MRI value p- 0.02 0.07 0.65 0.08 0.70 <0.001 , Mean Diffusivity; , Mean , WM hyperintensity. , WM hyperintensity. MD WMH Model 3 1.08 (1.01-1.15) 0.41 (0.16-1.07) 1.26 (0.46-3.43) 2.27 (0.91-5.68) 0.81 (0.28-2.36) NR NR NR NR NR NR 1.85 (1.29-2.65) 0.82 (0.71-0.93) 256.3 , hazard ratio; ratio; , hazard HR , white matter; matter; , white value WM p- <0.001 0.20 0.62 0.19 1.00 0.04 , grey matter; matter; , grey GM : standard deviation; deviation; : standard SD Model 2 1.10(1.04-1.16) 0.53 (0.20-1.39) 1.28 (0.48-3.38) 2.00 (0.71-5.67) 1.00 (0.35-2.81) 1.15 (1.01-1.32) NR NI NI NI NI NI 0.79 (0.68-0.90) 269.3 value p- <0.001 0.10 0.59 0.08 0.91 , diffusion tensor imaging; , diffusion DTI , not retained in the model; retained , not NR HR (95% CI) Model 1 1.12 (1.06-1.18) 0.45 (0.18-1.16) 1.31 (0.49-3.47) 2.39 (0.90-6.33) 1.06 (0.37-3.01) NI NI NI NI NI NI NI 0.78 (0.67-0.89) 274.9 : confidence interval; interval; : confidence CI /s 2 a mm -4 I, not included in the model; I, not N Akaike information criterion; criterion; information Akaike Characteristic (n = 494) Characteristic Demographics increase per 1-year Age, Male sex risk factors Vascular Smoking mellitus Diabetes Hypertension scores Clinical Gait speed, per 0.1m/s slower gait Cognitive index measures MRI and DTI mL increase, WMH volume, per 1-SD mL increase, WM volume, per 1-SD mL increase, GM volume, per 1-SD increase Lacunes, per No. x10 increase, MD of WM, per 1-SD (95% CI) C-statistic AIC Log transformed Log Table 5. Baseline indicator variables retained in Cox regression models for prediction of vascular mortality (n=26) mortality vascular of prediction for models regression in Cox retained variables 5. Baseline indicator Table AIC, Abbreviations: imaging; resonance magnetic a

106 FACTORS ASSOCIATED WITH 8-YEAR MORTALITY IN CEREBRAL SMALL VESSEL DISEASE.

Discussion This study shows that older age, lower gait speed, lower GM volumes and greater global MD of the WM at baseline significantly increased the 8-year risk of mortality in individuals 50 1 years and over with CSVD and together had the best predictive ability for all-cause mortality in our population. Cognitive performance and conventional MRI markers of CSVD (including WMH volume, WM volume and lacunes) did not significantly contribute to the prediction of mortality, although these factors were associated with 8-year mortality after adjustment for 2 age, sex, and vascular risk factors. This is a unique longitudinal study that assesses potential clinical and imaging risk factors of mortality in patients with CSVD. The strengths of this study include the large sample size, the complete follow-up for mortality, and its longitudinal and single-centre design, which 3 allowed us to consistently collect gait, cognitive, and imaging measures. Furthermore, all imaging data were analyzed with the reviewers blinded to clinical information with good intra and inter-rater variability.47 Some limitations need to be addressed. First, the observational design of the study prevents us 4 from elucidating the exact association between the baseline factors identified and mortality. However, it is conceivable that these factors reflect the vital health status of our participants rather than their having a direct causal association with mortality. Second, because age is the strongest predictor of mortality,147 the additional predictive value of gait speed, GM 5 volume, and global MD of the WM was relatively low. Nonetheless, risk factors or diseases underlying slower gait, lower GM volume, or lower WM microstructural integrity might be modifiable. Future research is needed to investigate whether treatment of these underlying risk factors results in improved survival. Third, our results on vascular mortality should be 6 interpreted with caution owing to limited statistical power. Fourth, we could not exclude the possibility that residual confounding by unmeasured variables (e.g., socioeconomic status and genetics) or years of uncontrolled vascular risk factors could, at least in part, have explained our results. Finally, although this was a hospital-based cohort study, we believe 7 that our results can be generalized to patients with CSVD referred to a general neurologic clinic because we included all consecutive patients with CSVD and there were no restrictions for admission to our department. 8 Of the clinical parameters included in our prediction model, gait speed was retained in all models and cognitive performance was not retained after including imaging markers. A possible explanation for this finding might be that gait relies not only on intact cerebral networks but also on the functioning of several organ systems, including respiratory, 9 circulatory, musculoskeletal, and peripheral nerve systems.25 As a result, slower gait might not be primarily caused by dysfunction of one system (e.g., the brain) but is probably due to accumulation of pathology among several organ systems, which is in accordance with 102 the results of one recent study. Because cognition is less affected by damage to other A organ systems compared with the brain, it could be that the association between cognitive performance and mortality is mediated by CSVD. Cognitive performance has been related

107 CHAPTER 6 to conventional markers of CSVD and DTI measures.148, 149 In our prediction model, cognitive index was not retained in the model when these imaging markers were added. Another interesting finding was that the microstructural integrity of the WM seems to be more important in the prediction of mortality in our population than the conventional MRI markers of CSVD because none of these markers were retained in the prediction model. An explanation for this finding might be that diffusion abnormalities in the WM probably better reflect the overall WM damage of our population since most of our participants had mild to moderate severe CSVD at baseline. Conventional MRI markers of CSVD probably reflect a small amount of WM abnormalities because it has been suggested that changes in diffusion parameters precede the development of WMH.91 Furthermore, other (neurodegenerative) abnormalities may have influenced the microstructural integrity of the WM of our participants and the association with mortality. GM volume has been associated with mortality in population-based studies.150 We showed that this factor had a significant contribution in the prediction of mortality in a CSVD population. As 17 (21.8%) of the patients who died had developed dementia during the follow-up of our study, this may be an explanation for the observed association between lower GM volume at baseline and mortality. However, the causes of increased mortality in patients with dementia have not been fully elucidated.151

Conclusions This study showed that, in the prediction of mortality in older adults with CSVD, older age, lower gait speed, lower GM volume, and a greater global MD of the WM were the factors primarily associated with 8-year all-cause mortality. These factors probably reflect the vital health status of this group. Future studies are needed to investigate the reproducibility of our prediction model on mortality. Furthermore, research is needed to elucidate the exact association between these risk factors and mortality and should investigate whether, for example, intervention in these risk factors (e.g., by treatment of CSVD and/or treatment of the underlying causes of gait impairment) could improve life expectancy in older adults with CSVD.

108 FACTORS ASSOCIATED WITH 8-YEAR MORTALITY IN CEREBRAL SMALL VESSEL DISEASE.

109

Part V Summary and discussion

7. General discussion and future perspectives CHAPTER 7

114 GENERAL DISCUSSION AND FUTURE PERSPECTIVES.

The overall objective of this thesis was to gain a better understanding of the brain changes underlying motor performance, including gait decline and incident parkinsonism, in cerebral small vessel disease (CSVD) using conventional magnetic resonance imaging (MRI) and 1 diffusion tensor imaging (DTI). Identification of determinants of adverse outcomes (i.e. mortality) in CSVD was another aim of this thesis. The studies described in this thesis are conducted as part of the RUN DMC study, a prospective cohort study, which investigates the risk factors and clinical consequences, including motor, cognitive and mood disturbances, of 2 brain changes as assessed by MRI and DTI. This cohort study consists of 503 participants with CSVD, aged between 50 and 85 years, who were initially free of parkinsonism and dementia. Baseline assessment (2006) included a cerebral MRI and extensive cognitive and motor evaluation, with, among others, quantitative measurement of gait and parkinsonian signs. 3 This assessment was repeated in 2011-2012 and in 2015 (9 years follow-up). In this chapter, I will discuss several important methodological considerations. Next, an overall view of the main findings of the studies in this thesis is provided. Finally, the clinical relevance of our findings and suggestions for future research are discussed. 4

Methodological considerations

Internal validity 5 The internal validity of a study refers to the reliability of the observed findings of the data collected. Internal validity depends on the extent to which potential sources of bias have been avoided.152 There are several forms of bias with selection bias, information bias and confounding being the most important. 6

Selection bias may occur when selection of participants is reliant on the outcome and has not been a random selection.152 In our study, selection bias may have occurred at both baseline and follow-up. At baseline, the initial response rate was 71.3%; non-participants 7 were older and had more severe white matter hyperintensities (WMH) measured by the Age- Related White Matter Changes (ARWMC) scale on a cerebral scan prior to inclusion.153 Higher age and higher load of WMH of this group may have resulted in more motor and cognitive 8 disabilities, which could have prevented them from participation. As a result, it is likely that we have missed patients with a higher degree of both motor impairment and CSVD. This bias might have led – if any - to an underestimation of the found associations between CSVD and the clinical outcomes, as well as to an underestimation of the incidence of parkinsonism and 9 mortality in our study population. A part of our study population (20.9%) did not participate in the follow-up assessment, because they deceased, were lost to follow-up or were not able to participate, leading to selective drop-outs (attrition bias). Since we were able to identify primary outcomes of almost all baseline participants, including on the presence A of parkinsonism and mortality, this bias had not influenced our results on these outcome measures. However, this bias might have influenced our analyses on gait decline, as from a

115 CHAPTER 7 substantial part (26.6%) of our initial participants gait assessment was not available at follow- up (chapter 3 and 4). In order to get insight in the potential magnitude of this attrition bias we compared the baseline characteristics between the participants and non-participants in our gait studies. We found that the non-participants were older, had a significant lower gait speed and higher load of CSVD at baseline; this may have resulted in an underestimation of the strength of the found associations.

Information bias might arise as a result of measurement errors in either the determinant (i.e. CSVD markers) or outcome (i.e. motor performance), which might lead to misclassification.152 The risk of measurement errors is higher when the determinant or outcome is not properly defined. A distinction can be made between differential misclassification, when the degree of misclassification is different between groups that are compared, and non-differential misclassification, indicating that the frequency of errors is approximately the same in groups being compared. In general, differential misclassification can lead to an under- or overestimation, whereas non-differential misclassification often leads to underestimation of the true association. The risk of information bias in the RUN DMC study was reduced by the single-centre design, which allowed us to systematically and uniformly collect information at baseline and follow-up. Furthermore, the assessment of the determinants (imaging markers) was performed independently and blinded to clinical information, minimizing differential misclassification. Information bias might however have occurred when determining imaging measures and motor performance, which will be discussed in the following section.

Measurement errors in MRI parameters: By using a 1.5-Tesla MRI with 5mm thick slices, instead of a 3.0-Tesla or even 7.0-Tesla MRI with small slices, underestimation might especially have occurred in the assessment of CSVD. Stronger magnetic field strengths with lower slice thickness, as well as susceptibility-weighted imaging have improved the detection of WMH, (micro)infarcts and microbleeds in recent years. It is however, unclear whether this improved detection is linearly present in the whole range of the CSVD spectrum or predominantly plays a role in those with a moderate to severe load of CSVD on 1.5-Tesla MRI. For brain volumetry we initially used Statistical Parametric Mapping 5 (SPM5) (chapter 2, 5 and 6) and later switched to a major update of the SPM software (SPM12) for the studies on gait decline (chapter 3 and 4). This recent updated version of SPM12 included, among others, a ‘New Segment’ segmentation procedure, which included additional tissue maps for non-brain soft-tissue (i.e. dura, venous sinus), bone and air/background. This might help to reduce the amount of non-brain tissue misclassified as grey matter or cerebrospinal fluid (CSF). Determining brain volumes with SPM5 might therefore have possibly led to an overestimation of grey matter volume and/or CSF in comparison to brain volumes determined with SPM12. In general, determining brain volumes with SPM in participants at the extremes of a spectrum (high brain volumes with slit like ventricles versus severe brain

116 GENERAL DISCUSSION AND FUTURE PERSPECTIVES. atrophy) is much more under the influence of measurement error than in participants with a relatively standard brain volume, with better registration to the probability distribution maps of SPM. This might have led to differential misclassification, because participants with 1 primary and secondary outcomes are expected to have a higher age and possibly more brain atrophy. To avoid the erroneous segmentation of WMH as grey matter, we corrected the T1 images for these segmentation errors using WMH mask. However, this procedure was not performed in SPM5, which may have led to an overestimation of grey matter, especially in 2 participants with a high WMH volume (differential misclassification). Since we found lower grey matter volume to be associated with incident vascular parkinsonism (chapter 5) and 8-year mortality (chapter 6), which were also associated with higher WMH volumes, this possible differential misclassification might not be of significance in our study. We tried to 3 reduce this bias by visual inspection of all scans by 1 rater. Corrections were made if co- registration errors, motion artefacts or segmentation failures had occurred. The microstructural integrity of the white matter was investigated with DTI by using a diffusion tensor model, which describes a Gaussian diffusion process.154 In complex brain tissue, i.e. 4 areas with crossing fibres and pathology, the water diffusion is non-Gaussian, which possibly have led to a misclassification of the water molecules diffusion in those areas.118 The DTI post processing technique we mainly used in this thesis was a Tract-Based Spatial Statistics (TBSS) analysis, whereby local fractional anisotropy maxima are projected onto a 5 skeleton of the core of major white matter tracts. This technique is easy to use, fast and is an automatic procedure, which reduces registration inaccuracies and smoothing procedure errors. Disadvantages of this technique may be that these maxima do not represent the true anatomical centre of a white matter tract in a diseased brain, particularly in the presence 6 of territorial infarcts, CSVD related lesions and/or atrophy, which could result in differential misclassification. We tried to overcome this, by excluding participants with territorial infarcts from the TBSS analyses. Since the white matter not included in the white matter tracts is not evaluated in TBSS, we included global DTI measures of the whole white matter in our 7 analyses as well.

Measurement errors in motor decline: The diagnosing of parkinsonism at follow-up in our 8 population was vulnerable to misclassification, as different case finding methods were used and not all diagnoses could be confirmed by neurological examination in our centre. For participants who were not able to visit our research centre, we had to rely on medical files and information from general practitioners. As parkinsonism is often accepted as 9 part of normal ageing and may not be recognized by general practitioners, it is likely that we have missed at least some incident cases and incorrectly classified them as having no parkinsonism, which probably resulted in an underestimation of our found associations (differential misclassification). Furthermore, the clinical distinction between different A subtypes of parkinsonism, is prone to errors, since the diagnostic accuracy is low in comparison to neuropathologic diagnosis.130 For the diagnosis of vascular parkinsonism (VP)

117 CHAPTER 7 the presence of cerebrovascular lesions on neuroimaging is required.21 This might have led to an overestimation of VP in our study, as all of our participants had at least some amount of CSVD and may have caused circular reasoning as well. In order to reduce the effects of possible misclassification of the different types of parkinsonism, we first performed the analyses with all types of parkinsonism, after which we performed separate analysis for a clinical diagnosis of VP, which showed similar results. Unfortunately, we were not able to take the course of disease into account or observe any treatment effect with Levodopa, which may have hampered the accuracy of diagnosis the subtypes of parkinsonism.

Confounding occurs in every observational study and may result in spurious associations. Confounding exists when an association between the determinant and outcome is explained by a third variable, which is not part of the causal pathway.152 There are several ways of controlling for confounding, including the use of multivariate analyses. We adjusted our analyses for major known confounders in the association between CSVD and motor decline, including age, gender and follow-up duration. We intentionally did not adjust for cardiovascular risk factors, as they are part of the causal chain of CSVD. We adjusted all our DTI analyses for traditional CSVD characteristics, which with no exception weakened the found associations. As structural integrity is now considered part of the CSVD spectrum, it can be questioned whether traditional CSVD markers are a true confounder and whether they are independent of these markers. We therefore presented DTI analyses in 2 models: without and with additional adjustment for CSVD. For our gait analyses, we excluded participants with other severe conditions than CSVD that affected gait, including severe arthritis, polyneuropathy, vision problems and cardiopulmonary diseases, because these diseases are not directly part of the causal pathway of CSVD. In addition, we chose other potential confounders (e.g. height, baseline gait parameters, and cognition) on information from previous studies and on our findings of univariate analyses. However, we could not rule out that residual confounding by unmeasured or unknown variables have influenced our associations.

Precision The precision of a study is high in the absence of random error, indicating fluctuations around the true value.152 Random error is inversely related to study size and is also related to the manner outcome measures are determined.152 To favour precision, we firstly included a large sample of older adults with CSVD at baseline and made a great effort to include as many of the baseline participants for follow-up examination. Second, we tried to minimize random errors in the assessment of determinants and outcomes by using the same research protocol at baseline and follow-up. We quantitatively assessed gait using GAITRite, which has a high test-retest reliability and validity93, 94 and averaged the gait parameters over two walks. Cognitive and other motor assessments, including unified Parkinson’s disease rating scale motor section (UPDRS-m) for the screening of parkinsonism, were performed by mostly

118 GENERAL DISCUSSION AND FUTURE PERSPECTIVES. the same and trained neurologist in training, promoting precision. Information on vital status was retrieved from the Dutch Municipal Personal Records database and the exact cause of death was obtained from the general practitioner, treating physician and medical files. 1 All imaging measures were also quantitatively assessed. For the traditional CSVD markers (manual rating of WMH, lacunes and microbleeds) an intra- and inter-rater agreement was obtained in a sample of 10% of the scans with good intra- and inter-rater reliabilities. The change of MRI-scanner between baseline and follow-up, leading to a distorting of the T1 2 images, has possibly affected precision (chapter 4). The effect of this scanner-upgrade is unknown, but previous studies have showed that brain volumes remained reliable; however, it may have introduced a bias in the main volume differences, which are probably random errors.111 We considered our longitudinal DTI measures as more robust, as the protocol did 3 not differ between these two time-points, except from the number of diffusion directions (30 versus 61 at baseline and follow-up), which has probably not influenced our results since we applied a diffusion tensor model with 6 degree of freedom. A previous study showed no differences in DTI parameters between scanners.112 The detection of WMH with an in-house 4 developed semi-automatic detection method used in chapter 4,155 yielded significant lower WMH volumes than with manual segmentation. This is possibly be explained by the fact that automatic procedures might minimize the false-positive classifications, because the signal intensity in WMH typically overlaps with that of normal tissue and the borders of abnormal- 5 normal tissue are difficult to notice with the human eye, promoting precision. Another advantage of an automatic segmentation is that it is completely reproducible; manual segmentation usually suffers from intra- and inter-observer variability, although the inter- rater variability for manual segmentation was good in our study. 6

Causal inference The RUN DMC study has a longitudinal design, which allowed us to investigate changes over time, regarding changes in imaging measures and clinical parameters, including gait decline 7 and the development of parkinsonism. This relative long follow-up period enabled us to study clinical significant changes in both CSVD imaging markers and clinical parameters. A temporal relationship strengthens the evidence for a link between these parameters in 8 comparison to cross-sectional studies. However, due to the observational nature of our study we were not able to study the directionality of the associations found in our studies. We assumed that CSVD leads to gait decline and parkinsonism. However, there is a possibility of reverse causality, indicating that gait decline or parkinsonism may lead to (progression of) 9 CSVD and loss of white matter microstructural integrity, e.g. by leading to motor inactivity and a sedentary lifestyle and thereby promoting cardiovascular risk factors.

119 CHAPTER 7

External validity The extent to which the results of a study can be generalized to a population beyond the sample studied is defined as external validity.152 The RUN DMC study includes a large group of older adults with a broad age range (50-85 years), with no dementia and parkinsonism at baseline, and with diverse neuroimaging characteristics of sporadic CSVD. The frequent coexistence of lacunes and microbleeds in our population, as well as the high prevalence of vascular risk factors at baseline suggests a vascular aetiology in most of our participants. Although our participants were seen in a university hospital, we feel that our results can be extended to a CSVD population aged ≥50 years referred to a general neurologic clinic, since we included all consecutive patients referred to our out-patient department with CSVD on neuroimaging. There were no restrictions for admission and participants were usually seen as a first opinion at our clinic. Furthermore, in community-dwelling populations the prevalence of vascular risk factors and CSVD-related lesions, with a prevalence of WMH of >90% in adults over 60 year of age9, is comparable to our study population. To get more insight into the generalization of our results we compared our cohort with other CSVD cohorts. One of the largest CSVD cohorts is the Leukoariosis and Disability in the elderly (LADIS) study (n=639), which is a multicentre and multinational (European) collected hospital-based cohort of CSVD patients.156 Participants were referred to the hospitals for similar reasons and the inclusion criteria greatly overlap with our study, except for age at inclusion; 65-84 year in the LADIS study (mean age (SD) 74.1 (5.0) years) versus 50-85 year in ours (65.7 (8.8) years). As a result, the LADIS study had a higher proportion of participants with Fazekas 3 score at baseline (25% versus 13% in our study). Furthermore, there was a female predominance in contrast to ours, which could also be the result of the higher age at baseline, as the female sex has a higher life expectancy. There was a similar distribution of vascular risk factors between the LADIS and the RUN DMC study. Therefore, we regard the RUN DMC cohort as a good representation of older patients (aged ≥ 50 years) with sporadic CSVD visiting a neurological clinic.

120 GENERAL DISCUSSION AND FUTURE PERSPECTIVES.

General discussion of the main findings

Motor decline in cerebral small vessel disease 1

Gait decline White matter atrophy and loss of white matter integrity were associated with gait decline in our CSVD population after 5 years of follow-up, by affecting stride length instead of cadence 2 (chapter 4). As previous studies have thoroughly established cross-sectional associations between CSVD and gait performance,17, 157 it was interesting to note that baseline CSVD was not associated with gait decline in our population (chapter 3), although we observed a considerable decline of gait (≥0.1m/s) after 5 years in more than 70% of our participants. 3 The selective drop-out, which is often inherent to a prospective cohort study, resulted in a relatively good study population with a lower baseline burden of CSVD and higher gait speed, which might have partially contributed to this result. Furthermore, our participants might be better able to compensate for CSVD pathology than the non-participants, as they 4 had higher white and grey matter volumes and had an overall better cognitive function at baseline as well. This hypothesis is captured in the reserve hypothesis, indicating that a reserve in brain or cognitive functioning can sustain more pathological damage, which might explain the lack of association between brain pathology and its clinical manifestation in 5 specific populations.158, 159 Recently, this concept has also been applied to motor function, by showing that older adults with a higher cognitive reserve (reflected by higher education) were less susceptible to the effect of WMH on gait speed, than persons with a low cognitive reserve.160 However, we did not investigate the existence of a possible brain or cognitive 6 reserve in our study population. Next, the multifactorial nature of gait decline could have prevented us from finding significant associations with CSVD at baseline. A recent study showed that accumulation of pathology in multiple organ systems was associated with gait decline; no specific organ system was found to be primarily or independently responsible for 7 gait decline.102 At last, by using a self-selected gait speed, instead of a maximum gait speed, the interindividual variations in gait speed between baseline and follow-up could have been smaller than with a challenging maximum gait speed161 in which compensation mechanisms 8 will possibly fail earlier. This may have led to underestimation of our associations. Only a few studies have investigated changes in both variables (imaging and gait measures) and their results are conflicting; one found that progression of WMH and white matter atrophy was associated with a greater decline in gait,104 which is in line with our results, while 9 others have found no associations.108 The associations between changes in DTI measures gait decline has not been studied before, but the results of the former study could imply that loss of white matter integrity might underlie the found association between white matter atrophy and gait decline. We observed that loss of white matter microstructural integrity is A associated with gait decline. We found the strongest associations between DTI measures and decline in stride length in the corpus callosum and corona radiata, which are important white

121 CHAPTER 7 matter tracts in motor control. The corpus callosum is not specific for motor functioning, but is also important in cognitive performance.114 Both CSVD162 and slowing of gait163 have been shown to be associated with cognitive decline in older adults. Since we did not adjust our analyses on gait decline for cognitive decline, we could not rule out the possibility that the observed association between corpus callosum integrity, as well as white matter atrophy, and gait decline is partially explained by cognitive decline. It has recently been suggested that the association between white matter damage and gait impairment is, at least in part, mediated by cognitive performance.164 Future studies should further unravel the underlying mechanisms of gait decline in CSVD, and investigate the interaction between gait and cognitive decline in CSVD as well.

Parkinsonism Until now, only retrospective and cross-sectional studies have investigated the relationship between CSVD and parkinsonism and yielded mostly contradictory results. We showed that CSVD at baseline, especially WHM volume and lacunes, is associated with the development of parkinsonism after 5 years (chapter 5). The development of VP was, in addition, associated with the presence of microbleeds and a lower grey matter volume at baseline, as well as with lower baseline structural integrity in multiple bifrontal white matter tracts. This latter finding is in accordance with the results of previous cross-sectional studies, which showed that DTI measures, especially of the frontal white matter can be of help in distinguishing healthy controls from patients with VP122 and also in differentiating among different other subtypes of parkinsonism.136 Our findings advocate a role of CSVD in the aetiology of parkinsonism and indicate that severity of CSVD, but possibly also the location of CSVD, in view of the lower bifrontal white matter integrity in our participants with VP, might be important in the development of parkinsonism. We did not investigate the role of location of the traditional CSVD markers in the development of parkinsonism. CSVD, especially WMH, in the frontal regions and lacunes in the basal ganglia and thalamus, has been associated with mild parkinsonian signs165, 166 and parkinsonism.21 We can propose several mechanisms how CSVD can contribute to the development of parkinsonism. First, CSVD could disrupt the basal ganglia-thalamo-cortical circuits, as well as other important (frontal) motor pathways, which might be an explanation why patients exhibit predominant gait problems. Our results might support this possible mechanism by finding lower structural integrity in motor white matter tracts in VP, including the internal capsule, corona radiata and the genu of the corpus callosum. Furthermore, we found a lower volume of grey matter (cortical and subcortical) structures in patients with VP, which might be caused by direct ischemic damage or by secondary neuronal degeneration to WMH, lacunes and/or microbleeds, although we made adjustments for these traditional CSVD markers. We could not rule out the possibility that other age-related neurodegenerative pathology could have caused lower grey matter volumes in our patients with VP as well. Second, accumulation of pathology, including CSVD, age-related nigro-striatal loss, as well as amyloid, tau and possible Lewy-body pathology,

122 GENERAL DISCUSSION AND FUTURE PERSPECTIVES. can lower the threshold for motor symptoms of parkinsonism to become apparent. This concept is strengthened by a previous study that showed that comorbid WMH in idiopathic Parkinson's disease were associated with more severe motor symptoms independent of the 1 degree of nigrostriatal dopaminergic denervation.167 Another indication of this mechanism might be the fact that our patients with parkinsonism were significantly older in comparison to those without parkinsonism with as a result probably more age-related pathology, apart from CSVD, although all our analyses were age-adjusted. This might possibly also explain 2 why parkinsonism is often a late feature in, relatively young, patients with cerebral autosomal dominant arteriopathy with subcortical infarcts (CADASIL).168 As our study did not include nuclear imaging, we were not able to unravel the contributions of these different pathologies to the development of parkinsonism in our cohort. 3 An interesting parallel of VP can be made with vascular dementia. There is a major overlap in symptoms, risk factors and neuroimaging characteristics between VP and vascular dementia.169 Gait difficulties are often reported in vascular dementia170 and vice versa.171 This might suggest that they are part of the same disease entity. However, why some patients 4 present with predominant gait problems, whereas others exhibit mainly cognitive problems is not exactly understood. Possible factors or mechanisms can be the location of CSVD markers and loss of white matter integrity in specific brain areas, or reduced structural cerebral network connectivity, which might capture the cumulative effects of CSVD as well 5 as concomitant neurodegenerative pathology172 (e.g. β-amyloid plaques, tau deposits and α-synuclein aggregates in the brain). Future research including these different factors may shed light on this issue. 6 Determinants of mortality in cerebral small vessel disease In the RUN DMC study, participants with the highest mortality risk were those with a high age, low gait speed, low grey matter volume and low structural integrity of the white matter (chapter 6). The most prominent finding in our study on 8-year mortality was that imaging 7 markers, including structural integrity of the white matter and grey matter volume, could improve the precision in prediction of mortality, next to demographic and clinical parameters (age and gait speed) in our CSVD population. DTI has been proven to be a possible surrogate 8 marker of adverse outcomes, including disability, cognitive disturbances and new onset strokes in CADASIL patients173 and functional outcome and mortality in community dwelling populations. 174-176 We showed that diffusion parameters had a higher predictive value for mortality than the traditional CSVD markers, probably because they better reflect the overall 9 damage to the white matter in CSVD.91 Another remarkable finding was that the presence of CSVD was associated with an increased risk of all-cause mortality, and not only with an increased risk of vascular death. This might be the result of the previous found association 102 between CSVD and pathologic changes in other organ systems. And this could also A be a reason why gait speed seemed to be a better predictor of mortality than cognitive performance, as cognition is less affected by damage to other organ systems than the brain.

123 CHAPTER 7

This is in line with a recent study in CADASIL patients (n=278), showing that the combination of demographic, clinical and MRI measures might be of help in the prediction of adverse outcomes after 3 years of follow-up, including incident stroke, severe disability, dementia or death.177 Gait difficulties, number of lacunes and total brain volume were found to be predictors of the combined endpoint, including all adverse outcomes mentioned above. DTI was not included in this study and the contribution of grey and white matter volume was not studied separately. Only age was found to be a predictor of 3-year mortality. However, this result should be interpreted with caution due to the small number of deaths (n=14).

Clinical implications Motor decline in CSVD has always been somewhat neglected in comparison to cognitive decline so far. Our results might be of help to physicians seeing a patient with CSVD who experience motor problems, including gait disturbances or parkinsonism. Our studies indicate that CSVD-related brain changes, especially brain atrophy and loss of white matter integrity should be considered as one of the possible causes in these motor problems. Furthermore, our results underpin the importance of examination of gait in patients with CSVD, with a special attention to stride length. Our results indicate that a decline in gait speed (by a decline in stride length) could be a reflection of progression of white matter pathology (chapter 4), might indicate early stage of parkinsonism (chapter 5), and might increase the risk of mortality (chapter 6). Therefore, it can be suggested that changes in gait speed need further evaluation in patients with CSVD. As it might reflect a lower vital health status, also the assessment of organ systems other than the brain should be incorporated in this evaluation, including cardiopulmonary and musculoskeletal systems. Possible cut-off points for this evaluation could be a gait speed decrease of ≥0.1m/s or the development of gait impairment (gait speed <1.0m/s), as these cut-off points have been consistently associated with adverse outcomes.25, 86 Gait speed, stride length and cadence can be easily obtained in the outpatient department by measuring the time and number of steps taken at a certain walking distance.

Surrogate imaging markers in CSVD Establishing MRI measures as a surrogate marker for clinical consequences of CSVD might be of great interest for several reasons. First, abnormalities on brain imaging often exist long before clinical symptoms occur and possibly might have a faster progression rate than clinical outcome measures. As a result, including surrogate imaging markers in trials can reduce sample size.178 Our results indicate that especially the microstructural integrity of the white matter and brain atrophy have the best potential to serve as a surrogate markers for motor decline, as well as for long-term mortality in populations with CSVD. A reason might be that brain atrophy and DTI measures better capture the overall damage of diverse pathologies in the brain. The structural integrity of the corpus callosum might be of special interest as a potential marker of deterioration of motor performance in CSVD, as we found lower baseline

124 GENERAL DISCUSSION AND FUTURE PERSPECTIVES. structural integrity in this white matter tract in patients with incident parkinsonism (chapter 5). In addition, we found that loss of structural integrity of the corpus callosum was also associated with gait decline (chapter 4). 1 Second, surrogate imaging markers might provide a better understanding of the mechanisms of clinical manifestations of CSVD, including deterioration in motor performance. This especially applies for DTI measures, since a number of pathological processes could underlie changes in DTI measures, including axonal degeneration and ischemic demyelination. Radial 2 diffusivity (RD) and axonal diffusivity (AD) have been proposed to give information on the type of neuronal damage; RD is proposed to give information on demyelination and AD is thought to represent axonal degeneration.116 In our CSVD population, an increase in MD and RD was associated with gait decline (chapter 4), which might advocate a role for ischemic 3 demyelination in CSVD related gait decline. This result should however, be interpreted with caution since diffusion parameters are dependent on eigenvalue sorting, which is difficult to reliably obtain in complex brain tissue.118 DTI can also be used to study network connectivity by using tractography, which might increase our understanding how CSVD can affect gait 4 and motor function from a network perspective. Studies using network analyses in motor decline are currently lacking. DTI is, however, not used in clinical practice yet and before these measures can be used as surrogate markers, future studies are needed to investigate the reproducibility of our results and study the standardization and validation of the DTI 5 protocol and its measures. Lastly, if surrogate markers are proven to have clinical relevance, they might be possible targets for intervention. To date, there are no established therapeutic interventions for prevention or treatment of CSVD.179 Some studies have shown that lowering blood pressure might reduce the progression of WMH.180, 181 Furthermore, in cross-sectional 6 studies it has been shown that physical activity may help to prevent or delay brain atrophy.182, 183 Randomized trials are necessary to investigate whether prevention of deterioration of these surrogate imaging markers e.g. by cardiovascular risk management or by increasing physical activity and ambulatory function can improve motor function, life-expectancy and cognitive 7 performance. The effect of cardiovascular risk management and exercise has already been studied in relation to cognitive decline and dementia, although the results are somewhat conflicting.184, 185 Until proven, it seems reasonable to promote a healthy lifestyle (e.g. quitting 8 smoking, promoting a balanced diet) and stimulate patients to be physical active in order to reduce potential risk factors for the development of brain pathology.

125 CHAPTER 7

Directions for future research With the ageing of the society, CSVD and its clinical consequences more and more burdens the health care system. CSVD is responsible for about 20% of all strokes worldwide186 and triples the risk of future stroke14 and doubles the risk of dementia.187 Future research is therefore needed, as little effective treatments for CSVD and cognitive and motor decline are available yet. Several directions for future research can be suggested. First, the results of our studies should be replicated and extended to other longitudinal studies and the hypotheses generated in this thesis should be tested. Future studies on the long- term prognosis of CSVD should ideally take functional dependence, besides mortality, into account, as hospitalization and being dependent on home care, have a significant impact on the perceived quality of life of older adults. A prediction model for functional dependence in CSVD increases the external validity towards clinical practice. Increasing knowledge on the determinants of a functional worse outcome in CSVD could possibly lead to the development of strategies for prevention or treatment options. In order to gain more insight into the pathogenesis of parkinsonism in CSVD cohorts and the contribution of diverse brain pathology (including CSVD, neurodegenerative pathology and normal ageing), prospective multimodal imaging, including MRI techniques and SPECT, should be incorporated in future studies. Second, we need a better understanding of the aetiology and progression of CSVD over time. As already mentioned, DTI is a promising tool for unravelling the underlying pathophysiology of white matter damage. In addition, the use of higher field strength imaging at 7 Tesla may improve our understanding by visualizing the vascular pathology itself in the perforating arteries and brain parenchym.188, 189 Furthermore, serial imaging with short intervals in- between, might provide a closer look at the progression of CSVD and changes of brain structure over time and give the opportunity to study the direct clinical consequences of these changes. A sub-study of the RUN DMC study is underway to further investigate this. It would also be interesting to investigate the development of CSVD and brain changes in younger healthy populations (e.g. aged 30 years) with and without vascular risk factors to study aetiology and brain changes and identify risk factors of these brain changes, including lifestyle and genetic factors. Another interesting direction for future research is to gain more insight into the mechanisms of the clinical consequences of CSVD by using DTI followed by tractography and network analyses, to study the contribution of disruption of white matter networks to clinical symptoms. It would also be of interest to study the difference and the interaction between cognitive and motor decline in CSVD populations with these techniques. Studying the longitudinal associations between these imaging measures and clinical measures would provide further evidence for a causal relationship.

126 GENERAL DISCUSSION AND FUTURE PERSPECTIVES.

127

8. Summary CHAPTER 8

130 SUMMARY.

Cerebral small vessel disease (CSVD), a disease of the small perforating cerebral arteries, arterioles, venules and capillaries, is very common in the elderly population, with a prevalence of >90% in individuals aged 60 years and over. On a cerebral magnetic resonance imaging 1 (MRI), this disease is visible as white matter hyperintensities (WMH), lacunes, microbleeds and brain atrophy. These imaging makers are considered as the traditional CSVD markers. Main risk factors are longstanding arterial hypertension, smoking and diabetes. CSVD is associated with a broad clinical spectrum, including cognitive disturbances, dementia, depression, 2 stroke, as well as motor problems, including gait disturbances and mild parkinsonian signs. It is assumed that these clinical symptoms are the result of disruption of white matter tracts and loss of connectivity between brain areas by CSVD. However, the traditional CSVD makers do not explain the whole clinical spectrum. Therefore, other factors, such as the underlying 3 white matter microstructural integrity, which can be assessed by diffusion tensor imaging (DTI), may play a role in the development of clinical symptoms in CSVD. In this thesis, I mainly focused on the brain changes in CSVD associated with the development of motor disturbances over time, including gait decline and the development of parkinsonism. 4 In addition, I studied the determinants of 8-year mortality in CSVD. All studies described in this thesis are conducted as part of the RUN DMC study, a prospective cohort study among 503 older adults, aged 50-85 years with CSVD on neuroimaging, which started in 2006. Follow- up assessment took place in 2011-2012. This study is designed to investigate the risk factors 5 and clinical consequences of brain changes assessed by conventional MRI and DTI. In this chapter I will summarize the main findings of each chapter.

Part II: Cerebral small vessel disease and cognitive performance 6 In chapter 2 we reported on the association between the white matter integrity, assessed by DTI, and cognitive performance, including verbal memory performance. Verbal memory failure may eventually result in cognitive decline and dementia in some. With a region of interest (ROI) approach and a Tract-Based Spatial Statistics (TBSS) analysis, we demonstrated 7 that the microstructural integrity of the cingulum is specifically associated with verbal memory performance at the cross-sectional level, independent of the presence of traditional CSVD markers. The cingulum is a white matter bundle, which connects the medial temporal 8 lobe structures (e.g. hippocampus) and the posterior cingulated cortex, and has a pivot role in memory function. In addition, we showed that microstructural integrity of the cingulum is associated with the structural integrity of the hippocampus. The cingular integrity was significantly lower in participants with a low structural integrity of the hippocampus 9 compared to those with a high structural integrity of the hippocampus. These results might indicate that cingular microstructural integrity could serve as a surrogate marker for (the development of) impaired verbal memory and possibly also dementia in individuals with CSVD. A

131 CHAPTER 8

Part III: Cerebral small vessel disease and motor performance This part of the thesis reports on the motor consequences of CSVD after 5 years of follow-up. Traditional CSVD markers (WMH volume, lacunes, microbleeds and white and grey matter volume), as well as DTI measures of the white matter were related to gait decline and the development of parkinsonism after 5 years. A TBSS analysis was used in each chapter, to study the microstructural integrity of important white matter tracts. In chapter 3 we investigated the associations between baseline CSVD and gait decline and the development of incident gait impairment, defined as a gait speed below 1.0m/s. We found no significant associations between baseline imaging markers and gait decline or incident gait impairment after 5 years in our population. In addition, the TBSS analysis revealed no significant differences in DTI measures between participants with and those without incident gait impairment after additional adjustments for the traditional CSVD markers. In chapter 4 we extended this study and investigated whether changes in these MRI and DTI markers are associated with gait decline after 5 years. We found that white matter atrophy and loss of white matter integrity were associated with gait decline. Gait decline was affected by a smaller stride length and not with a decline in cadence. Progression of the other markers of CSVD (WMH volume, lacunes and microbleeds) was not associated with gait decline. Changes in DTI measures, especially an increase in mean diffusivity and radial diffusivity, were associated with stride length decline. The strongest associations were found in the corpus callosum and posterior and anterior corona radiata and were independent of traditional CSVD markers. These findings suggest that progression of white matter pathology, especially white matter atrophy and loss of white matter structural integrity, should be considered as a possible cause of gait decline in older adults with CSVD. In chapter 5 we reported on the association between baseline CSVD and incident parkinsonism. Of our 503 baseline participants, parkinsonism developed in 20 participants after 5 years. The cumulative 5-year risk for parkinsonism was 3.5% (95% CI 1.9-5.2). We found that a high WMH volume and a high number of lacunes were associated with an increased 5-year risk of parkinsonism. For vascular parkinsonism, this risk was also increased by the presence of microbleeds and a low grey matter volume. Moreover, participants with vascular parkinsonism had a lower structural integrity in bifrontal white matter tracts in comparison to those without, independent of the presence of traditional CSVD markers. These findings are building the case for a role of CSVD in the aetiology of parkinsonism.

Part IV: Long-term mortality in cerebral small vessel disease In chapter 6 we studied the 8-year mortality in CSVD and investigated potential clinical and imaging factors associated with mortality. 80 Participants of the 503 died during the follow- up period (mean 7.8 years (SD 1.5)). All-cause mortality was highest in participants with the presence of lacunes and microbleeds, with the highest WMH volume and mean diffusivity of the white matter, and with the lowest white and grey matter volumes. In the prediction of mortality, older age, lower gait speed, lower grey matter volumes and greater mean diffusivity

132 SUMMARY. of the white matter were factors that best predicted 8-year mortality in our population with older adults with CSVD aged 50-85 years. Cognitive performance and other traditional CSVD markers were not retained in the prediction model. The predictive factors are probably a 1 reflection of the vital health status of our population.

Conclusion The studies described in this thesis showed that several brain changes are associated with 2 the development of motor disturbances after 5 years, as well as functional outcome in older adults with CSVD. The presence of traditional CSVD makers at baseline imaging increased the 5-year risk of incident parkinsonism, as well as the 8-year mortality. In addition, impaired white matter microstructural integrity at baseline was associated with incident vascular 3 parkinsonism, and was an important predictor of 8-year mortality, next to age, gait speed and grey matter volume in our population. The development of gait impairment or gait decline after 5 years was, however, not associated with baseline CSVD. Tough, gait decline was associated with white matter atrophy and loss of white matter microstructural integrity 4 after 5 years in our population. Our findings suggest that DTI holds promise for unravelling the underlying mechanisms, as well as for serving as a surrogate marker in the development of motor disturbances and functional outcome (e.g. mortality) in CSVD. More studies are needed to investigate the reproducibility 5 of our results and further elucidate the mechanisms of the development of clinical deficits in CSVD by using novel imaging techniques and network analysis. Furthermore, future research should also be directed at whether preventive strategies, e.g. on CSVD and gait decline, could improve motor and cognitive performance and functional outcome. 6

133

9. Summary in Dutch | Nederlandse samenvatting CHAPTER 9

136 SUMMARY IN DUTCH | NEDERLANDSE SAMENVATTING.

Met de Engelse term ‘cerebral small vessel disease’, afgekort CSVD, wordt schade aan de kleine bloedvaten in de hersenen aangeduid. Op een magnetic resonance imaging (MRI) scan van de hersenen is deze schade zichtbaar als witte stof afwijkingen, kleine herseninfarcten, ook 1 wel lacunaire infarcten genoemd, en microbloedingen, alsook verminderd hersenvolume; deze afwijkingen worden ook wel de traditionele CSVD markers genoemd. CSVD komt veel voor de op de oudere leeftijd, meer dan 90% van de mensen boven de 60 jaar heeft het in meer of mindere mate. Belangrijke risicofactoren voor CSVD zijn langer bestaande hoge 2 bloeddruk, roken en suikerziekte. CSVD is een belangrijke oorzaak voor problemen met het denken, dementie, depressie en een beroerte. Ook is CSVD geassocieerd met motorische problemen, zoals loopproblemen en milde parkinsonistische verschijnselen. Verondersteld wordt dat deze klachten ontstaan doordat deze aandoening leidt tot beschadigingen van 3 witte stof banen en daardoor leidt tot een verstoorde communicatie tussen verschillende hersengebieden die door witte stof banen met elkaar worden verbonden. De specifieke MRI afwijkingen correleren echter niet volledig met het klinisch spectrum van CSVD; personen met eenzelfde hoeveelheid beschadigingen op de MRI hoeven niet in dezelfde mate problemen 4 met bijvoorbeeld het geheugen of motoriek te ervaren. Met een speciale MRI techniek, de diffusion tensor imaging (DTI), kan de witte stof van de hersenen die verbindingsbanen bevat, in groter detail onderzocht worden. Kleine veranderingen in de witte stof, welke nog niet zichtbaar hoeven zijn op de gebruikelijke MRI, kunnen hiermee in kaart worden 5 gebracht. Deze techniek kan daardoor behulpzaam zijn om beter inzicht te krijgen in de hersenveranderingen die mogelijk ten grondslag liggen aan het ontstaan van klinische verschijnselen van CSVD. In dit proefschrift heb ik me vooral gericht op de hersenveranderingen bij ouderen met 6 CSVD die optreden bij achteruitgang in het motorisch functioneren, in het bijzonder bij het ontstaan van loopproblemen en parkinsonisme. Ook heb ik gekeken naar factoren die geassocieerd zijn met het 8-jaars sterfterisico in personen met CSVD, om een indruk te krijgen welke factoren een rol spelen bij een slechte uitkomst en mogelijk een afspiegeling zijn van 7 de gezondheidstoestand van deze groep. De onderzoeken in dit proefschrift zijn uitgevoerd in het kader van de Radboud University Nijmegen Diffusion tensor and Magnetic resonance imaging Cohort (RUN DMC) studie, 8 een prospectieve studie waarin 503 mensen tussen de 50-85 jaar oud en met tekenen van CSVD op de MRI-scan van de hersenen zijn ge-includeerd. Dit onderzoek is opgezet om de risicofactoren en de klinische gevolgen van hersenveranderingen in personen met CSVD te onderzoeken. Het onderzoek is gestart in 2006 en een eerste follow-up heeft plaatsgevonden 9 in 2011-2012. Tijdens deze 2 bezoeken ondergingen deelnemers een MRI-scan van de hersenen, kregen zij uitgebreide neuropsychologische testen om het geheugen in kaart te brengen en werd het lopen onderzocht met behulp van een elektronische loopmat, de GAITRite. De aanwezigheid van parkinsonisme werd onderzocht met behulp van de Unified A Parkinson’s Disease Rating Scale.

137 CHAPTER 9

Deel II: Cerebrale small vessel disease en cognitief functioneren In hoofdstuk 2 beschrijven we de associatie tussen de microstructurele integriteit van de witte stof, gemeten met DTI, en het verbale geheugen, onderdeel van het episodisch geheugen. Stoornissen in het verbale geheugen kunnen een voorstadium zijn van dementie. We hebben hiervoor 2 speciale DTI analyse methoden gebruikt, de zogenoemde region-of- interest (ROI) methode, waarbij op anatomische gronden een regio wordt gekozen om het gewenste baansysteem te selecteren, en een Tract-Based Spatial Statistics (TBSS) methode, waarmee gekeken kan worden naar de kern van de belangrijke witte stof banen. Met beide methoden vonden we dat de structurele integriteit van het cingulum, een witte stof baan die de mediale temporale structuren (zoals hippocampus) verbindt met de cingulaire cortex, cross-sectioneel geassocieerd is met het verbale geheugen. Deze associatie was onafhankelijk van de traditionele CSVD markers (witte stof afwijkingen, lacunes, microbleeds en hersenvolume). De structurele integriteit van het cingulum was ook geassocieerd met de integriteit van de hippocampus. De structurele integriteit van het cingulum bleek significant lager in deelnemers met een lagere hippocampale integriteit in vergelijking met deelnemers met een hogere hippocampale integriteit. Deze resultaten zouden er op kunnen wijzen dat de integriteit van het cingulum mogelijk gebruikt kan worden als marker voor (de ontwikkeling van) stoornissen in het verbale geheugen en daarmee een vroege marker van dementie kan zijn.

Deel III: Cerebrale small vessel disease en motorisch functioneren In deel 3 van dit proefschrift onderzochten we de gevolgen van CSVD op het motorisch functioneren na 5 jaar follow-up. We hebben ons hierbij gericht op het ontstaan van loopproblemen en de ontwikkeling van parkinsonisme. We hebben gekeken welke MRI markers geassocieerd zijn met deze achteruitgang in motorisch functioneren. Om de structurele integriteit van de witte stof banen te bestuderen maakten wij gebruik van de eerder genoemde TBSS methode. De associatie tussen CSVD op de baseline scan en de relatie tussen achteruitgang in het lopen en de ontwikkeling van een loopstoornis (gedefinieerd als een loopsnelheid <1.0m/s) 5 jaar later hebben we onderzocht in hoofdstuk 3. We stelden vast dat er geen relatie is tussen de hoeveelheid CSVD op de baseline scan en de achteruitgang in lopen in onze studie populatie. Ook liet de TBSS analyse geen verschillen zien in de microstructurele integriteit van de witte stof banen tussen personen met en zonder loopstoornissen, onafhankelijk van de traditionele CSVD markers. De baseline scan kan de ontwikkeling van loopproblemen dus niet goed voorspellen in onze populatie. Hoofdstuk 4 borduurt voort op het voorgaande hoofdstuk en hierin bestuderen we of hersenveranderingen op de MRI die optreden tussen baseline en follow-up onderzoek gerelateerd zijn aan de achteruitgang in het lopen. We laten zien dat zowel verlies van witte stof volume (witte stof atrofie) als een afname van de microstructurele integriteit van de witte stof geassocieerd zijn met achteruitgang in het lopen. Opmerkelijk was dat een toename van de hoeveelheid witte stof afwijkingen, lacunes

138 SUMMARY IN DUTCH | NEDERLANDSE SAMENVATTING. en microbleeds niet geassocieerd was met achteruitgang in het lopen. Een kleinere paslengte bleek de oorzaak voor de achteruitgang in het lopen en daarmee een lagere loopsnelheid, terwijl de cadans, het aantal stappen per minuut, gelijk bleef in onze populatie 5 jaar later. 1 Bij het bestuderen van de verschillende DTI parameters, laten we zien dat een toename van de gemiddelde diffusiviteit en de radiale diffusiviteit van witte stof banen, die wijzen op een slechtere microstructurele integriteit van de witte stof, de sterkste associaties tonen met een afname van de paslengte. De TBSS analyse laat zien dat vooral een verlies van structurele 2 integriteit in het corpus callosum (de hersenbalk) en het voorste deel van de corona radiata een belangrijke rol spelen in het ontstaan van loopproblemen in onze populatie. Onze resultaten suggereren dat veranderingen in de witte stof, zelfs voordat ze zichtbaar zijn op de gebruikelijke MRI, een rol spelen in het ontstaan van loopproblemen. 3 In hoofdstuk 5 beschrijven we de relatie tussen CSVD op de baseline scan en het ontstaan van parkinsonisme. Van de 503 deelnemers aan het baseline onderzoek, kregen 20 deelnemers de diagnose parkinsonisme 5 jaar later. Het 5-jaars cumulatieve risico op parkinsonisme was hiermee 3.5% (95% CI 1.9-5.2). In dit hoofdstuk laten we zien dat een 4 groter volume van witte stof afwijkingen en een hoger aantal lacunes geassocieerd zijn met een verhoogd 5-jaarsrisico op het ontwikkelen van parkinsonisme. Bij het alleen in ogenschouw nemen van de patiënten met vasculair parkinsonisme, wordt dit risico ook verhoogd door de aanwezigheid van microbleeds en een kleiner grijze stof volume. Verder 5 zagen we dat patiënten met vasculair parkinsonisme een lagere structurele integriteit van de bifrontale witte stof banen hebben in vergelijkingen met deelnemers zonder parkinsonisme, onafhankelijk van de traditionele CSVD markers. Deze bevindingen pleiten voor een rol van CSVD in het ontstaan van parkinsonisme. 6

Deel IV: Lange termijn sterfterisico in cerebrale small vessel disease In hoofdstuk 6 onderzochten we het 8-jaars sterfterisico bij mensen met CSVD en tevens onderzochten we potentiële klinische en beeldvormende factoren die geassocieerd zijn met 7 het 8-jaars sterfterisico. In totaal overleden 80 deelnemers van de 503 na een gemiddelde follow-up duur van 7.8 jaar (SD 1.5). Het 8-jaars sterfterisico is verhoogd voor deelnemers met de aanwezigheid van lacunes en microbleeds op de baseline MRI-scan, een verhoogde 8 gemiddelde diffusiviteit van de witte stof, alsook een groot volume van witte stof afwijkingen en een klein volume witte en grijze stof. De belangrijkste voorspellers voor het 8-jaars sterfterisico bleken een hogere leeftijd, lagere loopsnelheid, kleiner grijze stof volume en een verhoogde gemiddelde diffusiviteit van de witte stof. Deze factoren samen geven 9 waarschijnlijk een afspiegeling van de algemene gezondheidstoestand van onze populatie.

139 CHAPTER 9

Conclusie: De studies die beschreven zijn in dit proefschrift laten zien dat diverse hersenveranderingen geassocieerd zijn met het ontwikkelen van problemen op motorisch gebied en functionele uitkomst op lange termijn bij personen met CSVD. De aanwezigheid van traditionele CSVD makers op een MRI-scan verhoogt het 5-jaars risico op parkinsonisme, als ook het 8-jaars sterfterisico. Een lagere structurele integriteit van de witte stof is daarnaast geassocieerd met het ontstaan van vasculair parkinsonisme en is tevens een belangrijker voorspeller van mortaliteit in onze populatie, naast leeftijd, loopsnelheid en grijze stof volume. Het ontwikkelen van loopproblemen is niet gerelateerd aan de aanwezigheid van CSVD op de baseline MRI-scan, maar juist geassocieerd met het ontstaan van witte stof atrofie en verlies van microstructurele integriteit van de witte stof op de MRI-scan 5 jaar later. Uit onze resultaten blijkt dat DTI een belangrijke bijdrage kan leveren aan het inzichtelijk maken van de mechanismen die ten grondslag liggen aan het ontstaan van problemen in het motorisch functioneren en de functionele uitkomst (zoals mortaliteit) in personen met CSVD. Tevens doen onze resultaten vermoeden dat DTI parameters gebruikt kunnen worden als (vroege) marker voor ontstaan van problemen op deze gebieden. Toekomstige prospectieve studies zijn echter nodig om verder inzicht te verkrijgen in de onderliggende mechanismen van de klinische gevolgen van CSVD, onder andere ook door gebruik te maken van nieuwe beeldvormende technieken en netwerk analyse. Ook dient onderzocht te worden of het (preventief) behandelen van CSVD en loopproblemen, kan leiden tot vermindering van motorische en cognitieve problemen en verbetering van de functionele uitkomst.

140 SUMMARY IN DUTCH | NEDERLANDSE SAMENVATTING.

141

Part V| Appendices

A1 List of abbreviations PART VI

146 APPENDICES.

AD axial diffusivity AIC Akaike information criterion ANCOVA analysis of covariance 1 ARWMC age-related white matter changes scale CADASIL cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy CARASIL cerebral autosomal recessive arteriopathy with subcortical infarcts and 2 leukoencephalopathy CES-D centre of epidemiologic studies on depression scale CI confidence interval CSF cerebrospinal fluid 3 CSVD cerebral small vessel disease CT compute tomography DTI diffusion tensor imaging FA fractional anisotropy 4 FDR false discovery rate FLAIR fluid-attenuated inversion recovery FSL FMRIB’s software library GM grey matter 5 HR hazard ratio ICD-10 international statistical classification of disease and related health problems, 10th revision ICV intracranial volume 6 IPD idiopathic Parkinson’s disease IQR interquartile range MCI mild cognitive impairment MD mean diffusivity 7 MMSE Mini Mental State Examination MNI Montreal neurological institute MPRAGE magnetization-prepared rapid gradient-echo 8 MRI magnetic resonance imaging NAWM normal-appearing white matter NI not included NR not retained 9 NS not significant OPD outpatient department OR odds ratio PSP progressive supranuclear palsy A RAVLT Rey auditory verbal learning test RCFT Rey complex figure test

147 PART VI

RD radial diffusivity ROI region of interest RUN DMC Radboud university Nijmegen diffusion tensor and magnetic resonance imaging cohort SAT speed accuracy tradeoff SD standard deviation SPM statistical parametric mapping STRIVE standards for reporting vascular changes on neuroimaging TBSS Tract-Based Spatial Statistics TBV total brain volume TE echo time TI inversion time TIA transient ischemic attack TR time repetition UPDRS unified Parkinson’s disease rating scale UPDRS-m unified Parkinson’s disease rating scale motor section VIF variance inflation factor VP vascular parkinsonism WM white matter WMH white matter hyperintensities

148 APPENDICES.

149

A2 References PART VI

152 APPENDICES.

1. Grinberg LT, Thal DR. Vascular pathology in the aged human brain. Acta Neuropathol 2010;119:277-290. 2. Pantoni L. Cerebral small vessel disease: from pathogenesis and clinical characteristics to therapeutic challenges. Lancet Neurol 2010;9:689-701. 1 3. Mok V, Kim JS. Prevention and Management of Cerebral Small Vessel Disease. J Stroke 2015;17:111-122. 4. Grueter BE, Schulz UG. Age-related cerebral white matter disease (leukoaraiosis): a review. Postgrad Med J 2012;88:79-87. 5. Wardlaw JM, Smith EE, Biessels GJ, et al. Neuroimaging standards for research into small vessel 2 disease and its contribution to ageing and neurodegeneration. Lancet Neurol 2013;12:822-838. 6. Alexander AL, Lee JE, Lazar M, Field AS. Diffusion tensor imaging of the brain. Neurotherapeutics 2007;4:316-329. 7. O’Sullivan M, Summers PE, Jones DK, Jarosz JM, Williams SC, Markus HS. Normal-appearing white 3 matter in ischemic leukoaraiosis: a diffusion tensor MRI study. Neurology 2001;57:2307-2310. 8. Maniega SM, Valdes Hernandez MC, Clayden JD, et al. White matter hyperintensities and normal-appearing white matter integrity in the aging brain. Neurobiol Aging 2015;36:909-918. 9. de Leeuw FE, de Groot JC, Achten E, et al. Prevalence of cerebral white matter lesions in elderly 4 people: a population based magnetic resonance imaging study. The Rotterdam Scan Study. J Neurol Neurosurg Psychiatry 2001;70:9-14. 10. Das RR, Seshadri S, Beiser AS, et al. Prevalence and correlates of silent cerebral infarcts in the Framingham offspring study. Stroke 2008;39:2929-2935. 5 11. Vermeer SE, Koudstaal PJ, Oudkerk M, Hofman A, Breteler MM. Prevalence and risk factors of silent brain infarcts in the population-based Rotterdam Scan Study. Stroke 2002;33:21-25. 12. Vernooij MW, van der Lugt A, Ikram MA, et al. Prevalence and risk factors of cerebral microbleeds: the Rotterdam Scan Study. Neurology 2008;70:1208-1214. 6 13. Charlton RA, Barrick TR, McIntyre DJ, et al. White matter damage on diffusion tensor imaging correlates with age-related cognitive decline. Neurology 2006;66:217-222. 14. Debette S, Markus HS. The clinical importance of white matter hyperintensities on brain magnetic resonance imaging: systematic review and meta-analysis. Bmj 2010;341:c3666. 7 15. Rosso AL, Studenski SA, Chen WG, et al. Aging, the central nervous system, and mobility. J Gerontol A Biol Sci Med Sci 2013;68:1379-1386. 16. Zheng JJ, Delbaere K, Close JC, et al. White matter hyperintensities are an independent 8 predictor of physical decline in community-dwelling older people. Gerontology 2012;58:398-406. 17. de Laat KF, van Norden AG, Gons RA, et al. Gait in elderly with cerebral small vessel disease. Stroke 2010;41:1652-1658. 18. Louis ED, Luchsinger JA. History of vascular disease and mild parkinsonian signs in community- 9 dwelling elderly individuals. Arch Neurol 2006;63:717-722. 19. Murray ME, Senjem ML, Petersen RC, et al. Functional impact of white matter hyperintensities in cognitively normal elderly subjects. Arch Neurol 2010;67:1379-1385. 20. Yamanouchi H, Nagura H. Neurological signs and frontal white matter lesions in vascular A parkinsonism. A clinicopathologic study. Stroke 1997;28:965-969.

153 PART VI

21. Zijlmans JC, Daniel SE, Hughes AJ, Revesz T, Lees AJ. Clinicopathological investigation of vascular parkinsonism, including clinical criteria for diagnosis. Mov Disord 2004;19:630-640. 22. Kerber KA, Whitman GT, Brown DL, Baloh RW. Increased risk of death in community-dwelling older people with white matter hyperintensities on MRI. J Neurol Sci 2006;250:33-38. 23. Ikram MA, Vernooij MW, Vrooman HA, Hofman A, Breteler MM. Brain tissue volumes and small vessel disease in relation to the risk of mortality. Neurobiol Aging 2009;30:450-456. 24. Sachs GA, Carter R, Holtz LR, et al. Cognitive impairment: an independent predictor of excess mortality: a cohort study. Ann Intern Med 2011;155:300-308. 25. Studenski S, Perera S, Patel K, et al. Gait speed and survival in older adults. Jama 2011;305:50-58. 26. de Groot JC, de Leeuw FE, Oudkerk M, et al. Cerebral white matter lesions and cognitive function: the Rotterdam Scan Study. Ann Neurol 2000;47:145-151. 27. De Groot JC, De Leeuw FE, Oudkerk M, et al. Periventricular cerebral white matter lesions predict rate of cognitive decline. Ann Neurol 2002;52:335-341. 28. Vermeer SE, Prins ND, den Heijer T, Hofman A, Koudstaal PJ, Breteler MM. Silent brain infarcts and the risk of dementia and cognitive decline. N Engl J Med 2003;348:1215-1222. 29. Brun A, Englund E. A white matter disorder in dementia of the Alzheimer type: a pathoanatomical study. Ann Neurol 1986;19:253-262. 30. Gouw AA, Seewann A, Vrenken H, et al. Heterogeneity of white matter hyperintensities in Alzheimer’s disease: post-mortem quantitative MRI and neuropathology. Brain 2008;131:3286-3298. 31. Acosta-Cabronero J, Williams GB, Pengas G, Nestor PJ. Absolute diffusivities define the landscape of white matter degeneration in Alzheimer’s disease. Brain 2010;133:529-539. 32. Basser PJ, Mattiello J, LeBihan D. Estimation of the effective self-diffusion tensor from the NMR spin echo. J Magn Reson B 1994;103:247-254. 33. Pierpaoli C, Jezzard P, Basser PJ, Barnett A, Di Chiro G. Diffusion tensor MR imaging of the human brain. Radiology 1996;201:637-648. 34. Kantarci K, Petersen RC, Boeve BF, et al. DWI predicts future progression to Alzheimer disease in amnestic mild cognitive impairment. Neurology 2005;64:902-904. 35. Muller MJ, Greverus D, Weibrich C, et al. Diagnostic utility of hippocampal size and mean diffusivity in amnestic MCI. Neurobiol Aging 2007;28:398-403. 36. Fellgiebel A, Muller MJ, Wille P, et al. Color-coded diffusion-tensor-imaging of posterior cingulate fiber tracts in mild cognitive impairment. Neurobiol Aging 2005;26:1193-1198. 37. Sepulcre J, Masdeu JC, Sastre-Garriga J, et al. Mapping the brain pathways of declarative verbal memory: Evidence from white matter lesions in the living human brain. Neuroimage 2008;42:1237-1243. 38. Sexton CE, Mackay CE, Lonie JA, et al. MRI correlates of episodic memory in Alzheimer’s disease, mild cognitive impairment, and healthy aging. Psychiatry Res 2010;184:57-62. 39. Catheline G, Periot O, Amirault M, et al. Distinctive alterations of the cingulum bundle during aging and Alzheimer’s disease. Neurobiol Aging 2010;31:1582-1592. 40. Choo IH, Lee DY, Oh JS, et al. Posterior cingulate cortex atrophy and regional cingulum disruption in mild cognitive impairment and Alzheimer’s disease. Neurobiol Aging 2010;31:772-779.

154 APPENDICES.

41. Chua TC, Wen W, Chen X, et al. Diffusion tensor imaging of the posterior cingulate is a useful biomarker of mild cognitive impairment. Am J Geriatr Psychiatry 2009;17:602-613. 42. Kiuchi K, Morikawa M, Taoka T, et al. Abnormalities of the uncinate fasciculus and posterior 1 cingulate fasciculus in mild cognitive impairment and early Alzheimer’s disease: a diffusion tensor tractography study. Brain Res 2009;1287:184-191. 43. Nakata Y, Sato N, Nemoto K, et al. Diffusion abnormality in the posterior cingulum and hippocampal volume: correlation with disease progression in Alzheimer’s disease. Magn Reson Imaging 2 2009;27:347-354. 44. Xie S, Xiao JX, Wang YH, Wu HK, Gong GL, Jiang XX. Evaluation of bilateral cingulum with tractography in patients with Alzheimer’s disease. Neuroreport 2005;16:1275-1278. 45. Zhang Y, Schuff N, Jahng GH, et al. Diffusion tensor imaging of cingulum fibers in mild cognitive 3 impairment and Alzheimer disease. Neurology 2007;68:13-19. 46. Smith SM, Jenkinson M, Johansen-Berg H, et al. Tract-based spatial statistics: voxelwise analysis of multi-subject diffusion data. Neuroimage 2006;31:1487-1505. 47. van Norden AG, de Laat KF, Gons RA, et al. Causes and consequences of cerebral small vessel disease. 4 The RUN DMC study: a prospective cohort study. Study rationale and protocol. BMC Neurol 2011;11:29. 48. Erkinjuntti T. Subcortical vascular dementia. Cerebrovasc Dis 2002;2:58-60. 49. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. Washington DC 2000. 5 50. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189-198. 51. Van der Elst W, van Boxtel MP, van Breukelen GJ, Jolles J. Rey’s verbal learning test: normative data for 1855 healthy participants aged 24-81 years and the influence of age, sex, education, and mode 6 of presentation. J Int Neuropsychol Soc 2005;11:290-302. 52. Houx PJ, Jolles J, Vreeling FW. Stroop interference: aging effects assessed with the Stroop Color- Word Test. Exp Aging Res 1993;19:209-224. 53. Stroop JR. Studies of interference in serial verbal reactions. J Exp Psychol 1935;18:643–662. 7 54. Sternberg S. Memory-scanning: mental processes revealed by reaction-time experiments. Am Sci 1969;57:421-457. 55. Lezak MD. Neuropsychological Assessment. New York: Oxford University Press, 1976. 8 56. Osterrieth PA. Le test de kopie dune figure complexe. Contribution a l’étude de la perception et de la mémoire. Arch Psychol 1944;30:206-353. 57. van Norden AG, de Laat KF, Fick I, et al. Diffusion tensor imaging of the hippocampus and verbal memory performance: the RUN DMC study. Hum Brain Mapp 2012;33:542-551. 9 58. Vakil E, Blachstein H. Rey Auditory-Verbal Learning Test: structure analysis. J Clin Psychol 1993;49:883-890. 59. van Norden AG, de Laat KF, van Dijk EJ, et al. Diffusion tensor imaging and cognition in cerebral small vessel disease: the RUN DMC study. Biochim Biophys Acta 2012;1822:401-407. A 60. Yushkevich PA, Piven J, Hazlett HC, et al. User-guided 3D active contour segmentation of anatomical structures: significantly improved efficiency and reliability. Neuroimage 2006;31:1116-1128.

155 PART VI

61. Duvernoy H. The Human Hippocampus: Functional Anatomy, Vascularization and Serial Sections with MRI (hardcover). New York: Springer-Verlag, 1997. 62. Mai JP, G.; Voss, T. Atlas of the Human Brain. San Diego, CA: Elsevier, 2007. 63. Geuze E, Vermetten E, Bremner JD. MR-based in vivo hippocampal volumetrics: 2. Findings in neuropsychiatric disorders. Mol Psychiatry 2005;10:160-184. 64. Ashburner J, Friston KJ. Unified segmentation. Neuroimage 2005;26:839-851. 65. Zwiers MP. Patching cardiac and head motion artefacts in diffusion-weighted images. Neuroimage 2010;53:565-575. 66. Andersson JL, Skare S. A model-based method for retrospective correction of geometric distortions in diffusion-weighted EPI. Neuroimage 2002;16:177-199. 67. Smith SM, Jenkinson M, Woolrich MW, et al. Advances in functional and structural MR image analysis and implementation as FSL. Neuroimage 2004;23 Suppl 1:S208-219. 68. Eickhoff SB, Stephan KE, Mohlberg H, et al. A new SPM toolbox for combining probabilistic cytoarchitectonic maps and functional imaging data. Neuroimage 2005;25:1325-1335. 69. Schmahmann JD, Pandya DN. Fiber Pathways of the Brain. New York: Oxford University Press, 2006. 70. Hochstenbach J, Mulder T, van Limbeek J, Donders R, Schoonderwaldt H. Cognitive decline following stroke: a comprehensive study of cognitive decline following stroke. J Clin Exp Neuropsychol 1998;20:503-517. 71. Radloff LS. The CES-D Scale: a self-report depression scale for research in the general population. Appl Psychol Meas 1977;1:385-401. 72. Nichols TE, Holmes AP. Nonparametric permutation tests for functional neuroimaging: a primer with examples. Hum Brain Mapp 2002;15:1-25. 73. Smith SM, Nichols TE. Threshold-free cluster enhancement: addressing problems of smoothing, threshold dependence and localisation in cluster inference. Neuroimage 2009;44:83-98. 74. Bozzali M, Giulietti G, Basile B, et al. Damage to the cingulum contributes to Alzheimer’s disease pathophysiology by deafferentation mechanism. Hum Brain Mapp 2012;33:1295-1308. 75. Kantarci K, Senjem ML, Avula R, et al. Diffusion tensor imaging and cognitive function in older adults with no dementia. Neurology 2011;77:26-34. 76. Mielke MM, Kozauer NA, Chan KC, et al. Regionally-specific diffusion tensor imaging in mild cognitive impairment and Alzheimer’s disease. Neuroimage 2009;46:47-55. 77. Coleman M. Axon degeneration mechanisms: commonality amid diversity. Nat Rev Neurosci 2005;6:889-898. 78. Bartzokis G. Age-related myelin breakdown: a developmental model of cognitive decline and Alzheimer’s disease. Neurobiol Aging 2004;25:5-18; author reply 49-62. 79. Reisberg B, Franssen EH, Souren LE, Auer SR, Akram I, Kenowsky S. Evidence and mechanisms of retrogenesis in Alzheimer’s and other dementias: management and treatment import. Am J Alzheimers Dis Other Demen 2002;17:202-212. 80. Englund E, Brun A. White matter changes in dementia of Alzheimer’s type: the difference in vulnerability between cell compartments. Histopathology 1990;16:433-439.

156 APPENDICES.

81. Shim YS, Yoon B, Shon YM, Ahn KJ, Yang DW. Difference of the hippocampal and white matter microalterations in MCI patients according to the severity of subcortical vascular changes: neuropsychological correlates of diffusion tensor imaging. Clin Neurol Neurosurg 2008;110:552-561. 1 82. Head D, Buckner RL, Shimony JS, et al. Differential vulnerability of anterior white matter in nondemented aging with minimal acceleration in dementia of the Alzheimer type: evidence from diffusion tensor imaging. Cereb Cortex 2004;14:410-423. 83. Christman SD, Propper RE. Superior episodic memory is associated with interhemispheric 2 processing. Neuropsychology 2001;15:607-616. 84. Madden DJ, Turkington TG, Provenzale JM, et al. Adult age differences in the functional neuroanatomy of verbal recognition memory. Hum Brain Mapp 1999;7:115-135. 85. Abellan van Kan G, Rolland Y, Andrieu S, et al. Gait speed at usual pace as a predictor of adverse 3 outcomes in community-dwelling older people an International Academy on Nutrition and Aging (IANA) Task Force. J Nutr Health Aging 2009;13:881-889. 86. Cesari M, Kritchevsky SB, Penninx BW, et al. Prognostic value of usual gait speed in well-functioning older people--results from the Health, Aging and Body Composition Study. J Am Geriatr Soc 4 2005;53:1675-1680. 87. Aribisala BS, Gow AJ, Bastin ME, et al. Associations between level and change in physical function and brain volumes. PLoS One 2013;8:e80386. 88. Soumare A, Elbaz A, Zhu Y, et al. White matter lesions volume and motor performances in the 5 elderly. Ann Neurol 2009;65:706-715. 89. Rosano C, Kuller LH, Chung H, Arnold AM, Longstreth WT, Jr., Newman AB. Subclinical brain magnetic resonance imaging abnormalities predict physical functional decline in high-functioning older adults. J Am Geriatr Soc 2005;53:649-654. 6 90. Silbert LC, Nelson C, Howieson DB, Moore MM, Kaye JA. Impact of white matter hyperintensity volume progression on rate of cognitive and motor decline. Neurology 2008;71:108-113. 91. de Groot M, Verhaaren BF, de Boer R, et al. Changes in normal-appearing white matter precede development of white matter lesions. Stroke 2013;44:1037-1042. 7 92. de Laat KF, Tuladhar AM, van Norden AG, Norris DG, Zwiers MP, de Leeuw FE. Loss of white matter integrity is associated with gait disorders in cerebral small vessel disease. Brain 2011;134:73-83. 93. Menz HB, Latt MD, Tiedemann A, Mun San Kwan M, Lord SR. Reliability of the GAITRite walkway 8 system for the quantification of temporo-spatial parameters of gait in young and older people. Gait Posture 2004;20:20-25. 94. Bilney B, Morris M, Webster K. Concurrent related validity of the GAITRite walkway system for quantification of the spatial and temporal parameters of gait. Gait Posture 2003;17:68-74. 9 95. Willey JZ, Scarmeas N, Provenzano FA, Luchsinger JA, Mayeux R, Brickman AM. White matter hyperintensity volume and impaired mobility among older adults. J Neurol 2013;260:884-890. 96. Maldjian JA, Laurienti PJ, Kraft RA, Burdette JH. An automated method for neuroanatomic and cytoarchitectonic atlas-based interrogation of fMRI data sets. Neuroimage 2003;19:1233-1239. A 97. van der Holst HM, van Uden IW, Tuladhar AM, et al. Cerebral small vessel disease and incident parkinsonism: The RUN DMC study. Neurology 2015;85:1569-1577.

157 PART VI

98. Colliot O, Chetelat G, Chupin M, et al. Discrimination between Alzheimer disease, mild cognitive impairment, and normal aging by using automated segmentation of the hippocampus. Radiology 2008;248:194-201. 99. Basser PJ, Jones DK. Diffusion-tensor MRI: theory, experimental design and data analysis - a technical review. NMR Biomed 2002;15:456-467. 100. van Uden IWM, van der Holst HM, Schaapsmeerders P, et al. Baseline white matter microstructural integrity is not related to cognitive decline after 5 years: The RUN DMC study. BBA Clinical 2015;4:108-114. 101. Mahoney FI, Barthel DW. Functional Evaluation: The Barthel Index. Md State Med J 1965;14:61-65. 102. Rosso AL, Sanders JL, Arnold AM, et al. Multisystem physiologic impairments and changes in gait speed of older adults. J Gerontol A Biol Sci Med Sci 2015;70:319-324. 103. Starr JM, Leaper SA, Murray AD, et al. Brain white matter lesions detected by magnetic resonance [correction of resosnance] imaging are associated with balance and gait speed. J Neurol Neurosurg Psychiatry 2003;74:94-98. 104. Callisaya ML, Beare R, Phan TG, et al. Brain structural change and gait decline: a longitudinal population-based study. J Am Geriatr Soc 2013;61:1074-1079. 105. Zeestraten EA, Benjamin P, Lambert C, et al. Application of Diffusion Tensor Imaging Parameters to Detect Change in Longitudinal Studies in Cerebral Small Vessel Disease. PLoS One 2016;11:e0147836. 106. Srikanth V, Phan TG, Chen J, Beare R, Stapleton JM, Reutens DC. The location of white matter lesions and gait--a voxel-based study. Ann Neurol 2010;67:265-269. 107. Van der Holst HM, Van Uden IW, De Laat KF, et al. Baseline cerebral small vessel disease is not associated with gait decline after 5 years. Mov disord clin practice 2016 Nov.;10.1002/mdc3.12435. 108. Moscufo N, Wolfson L, Meier D, et al. Mobility decline in the elderly relates to lesion accrual in the splenium of the corpus callosum. Age 2012;34:405-414. 109. Ghafoorian M, Karssemeijer N, van Uden IW, et al. Automated detection of white matter hyperintensities of all sizes in cerebral small vessel disease. Med Phys 2016;43:6246. 110. Genovese CR, Lazar NA, Nichols T. Thresholding of statistical maps in functional neuroimaging using the false discovery rate. Neuroimage 2002;15:870-878. 111. Jovicich J, Czanner S, Han X, et al. MRI-derived measurements of human subcortical, ventricular and intracranial brain volumes: Reliability effects of scan sessions, acquisition sequences, data analyses, scanner upgrade, scanner vendors and field strengths. Neuroimage 2009;46:177-192. 112. Huang L, Wang X, Baliki MN, Wang L, Apkarian AV, Parrish TB. Reproducibility of structural, resting-state BOLD and DTI data between identical scanners. PLoS One 2012;7:e47684. 113. Rosario BL, Rosso AL, Aizenstein HJ, et al. Cerebral White Matter and Slow Gait: Contribution of Hyperintensities and Normal-appearing Parenchyma. J Gerontol A Biol Sci Med Sci 2016. 114. Chao YP, Cho KH, Yeh CH, Chou KH, Chen JH, Lin CP. Probabilistic topography of human corpus callosum using cytoarchitectural parcellation and high angular resolution diffusion imaging tractography. Hum Brain Mapp 2009;30:3172-3187. 115. Jang SH. A review of corticospinal tract location at corona radiata and posterior limb of the internal capsule in human brain. NeuroRehabilitation 2009;24:279-283.

158 APPENDICES.

116. Song SK, Sun SW, Ju WK, Lin SJ, Cross AH, Neufeld AH. Diffusion tensor imaging detects and differentiates axon and myelin degeneration in mouse optic nerve after retinal ischemia. Neuroimage 2003;20:1714-1722. 1 117. Englund E. Neuropathology of white matter lesions in vascular cognitive impairment. Cerebrovasc Dis 2002;13 Suppl 2:11-15. 118. Wheeler-Kingshott CA, Cercignani M. About “axial” and “radial” diffusivities. Magn Reson Med 2009;61:1255-1260. 2 119. Baezner H, Blahak C, Poggesi A, et al. Association of gait and balance disorders with age-related white matter changes: the LADIS study. Neurology 2008;70:935-942. 120. Kreisel SH, Blahak C, Bazner H, et al. Deterioration of gait and balance over time: the effects of age-related white matter change--the LADIS study. Cerebrovasc Dis 2013;35:544-553. 3 121. Cochrane CJ, Ebmeier KP. Diffusion tensor imaging in parkinsonian syndromes: a systematic review and meta-analysis. Neurology 2013;80:857-864. 122. Wang HC, Hsu JL, Leemans A. Diffusion tensor imaging of vascular parkinsonism: structural changes in cerebral white matter and the association with clinical severity. Arch Neurol 2012;69:1340-1348. 4 123. Roman GC, Erkinjuntti T, Wallin A, Pantoni L, Chui HC. Subcortical ischaemic vascular dementia. Lancet Neurol 2002;1:426-436. 124. Fahn S MC, Calne D, Goldstein M, editors. Recent developments in Parkinson’s disease: Florham Park, NJ. MacMillan Healthcare Information, 1987. 5 125. Litvan I, Bhatia KP, Burn DJ, et al. Movement Disorders Society Scientific Issues Committee report: SIC Task Force appraisal of clinical diagnostic criteria for Parkinsonian disorders: Mov Disord. 2003 May;18(5):467-86., 2003. 126. Buchman AS, Shulman JM, Nag S, et al. Nigral pathology and parkinsonian signs in elders without 6 Parkinson disease. Ann Neurol 2012;71:258-266. 127. Hughes AJ, Daniel SE, Kilford L, Lees AJ. Accuracy of clinical diagnosis of idiopathic Parkinson’s disease: a clinico-pathological study of 100 cases. J Neurol Neurosurg Psychiatry 1992;55:181-184. 128. Fazekas F, Chawluk JB, Alavi A, Hurtig HI, Zimmerman RA. MR signal abnormalities at 1.5 7 T in Alzheimer’s dementia and normal aging. AJR Am J Roentgenol 1987;149:351-356. 129. de Lau LM, Giesbergen PC, de Rijk MC, Hofman A, Koudstaal PJ, Breteler MM. Incidence of parkinsonism and Parkinson disease in a general population: the Rotterdam Study. Neurology 8 2004;63:1240-1244. 130. Horvath J, Burkhard PR, Bouras C, Kovari E. Etiologies of Parkinsonism in a century-long autopsy- based cohort. Brain Pathol 2013;23:28-33. 131. Stern MB, Braffman BH, Skolnick BE, Hurtig HI, Grossman RI. Magnetic resonance imaging 9 in Parkinson’s disease and parkinsonian syndromes. Neurology 1989;39:1524-1526. 132. Piccini P, Pavese N, Canapicchi R, et al. White matter hyperintensities in Parkinson’s disease. Clinical correlations. Arch Neurol 1995;52:191-194. 133. Schwartz RS, Halliday GM, Cordato DJ, Kril JJ. Small-vessel disease in patients with Parkinson’s A disease: a clinicopathological study. Mov Disord 2012;27:1506-1512.

159 PART VI

134. Dalaker TO, Larsen JP, Bergsland N, et al. Brain atrophy and white matter hyperintensities in early Parkinson’s disease(a). Mov Disord 2009;24:2233-2241. 135. Meijer FJ, Bloem BR, Mahlknecht P, Seppi K, Goraj B. Update on diffusion MRI in Parkinson’s disease and atypical parkinsonism. J Neurol Sci 2013;332:21-29. 136. Deverdun J, Menjot de Champfleur S, Cabello-Aguilar S, et al. Diffusion tensor imaging differentiates vascular parkinsonism from parkinsonian syndromes of degenerative origin in elderly subjects. Eur J Radiol 2014;83:2074-2079. 137. Galvan A, Wichmann T. Pathophysiology of parkinsonism. Clin Neurophysiol 2008;119:1459-1474. 138. Umegaki H, Roth GS, Ingram DK. Aging of the striatum: mechanisms and interventions. Age 2008;30:251-261. 139. Inzitari D, Pracucci G, Poggesi A, et al. Changes in white matter as determinant of global functional decline in older independent outpatients: three year follow-up of LADIS (leukoaraiosis and disability) study cohort. BMJ 2009;6. 140. Perna L, Wahl HW, Mons U, Saum KU, Holleczek B, Brenner H. Cognitive impairment, all- cause and cause-specific mortality among non-demented older adults. Age Ageing 2014;2. 141. Conijn MM, Kloppenborg RP, Algra A, et al. Cerebral small vessel disease and risk of death, ischemic stroke, and cardiac complications in patients with atherosclerotic disease: the Second Manifestations of ARTerial disease-Magnetic Resonance (SMART-MR) study. Stroke 2011;42:3105-3109. 142. International Statistical Classification of Diseases and Related Health Problems. Geneva, Switzerland: World Health Organization, 1992. 143. Rosendorff C, Black HR, Cannon CP, et al. Treatment of hypertension in the prevention and management of ischemic heart disease: a scientific statement from the American Heart Association Council for High Blood Pressure Research and the Councils on Clinical Cardiology and Epidemiology and Prevention. Circulation 2007;115:2761-2788. 144. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing riks. J Am Stat Assoc 1999;94:496-509. 145. Akaike H. A new look at the statistical model identification. IEEE Transactions on Automatic Control 1974;AC-19:716-723. 146. Burnham K, Anderson D, Huyvaert K. AIC model selection and multimodel inference in behavioral ecology: some background, observations, and comparisons. Behavioral Ecology and Sociobiology 2011;65:23-35. 147. Walter S, Mackenbach J, Voko Z, et al. Genetic, physiological, and lifestyle predictors of mortality in the general population. Am J Public Health 2012;102:16. 148. van Norden AG, van Uden IW, de Laat KF, van Dijk EJ, de Leeuw FE. Cognitive function in small vessel disease: the additional value of diffusion tensor imaging to conventional magnetic resonance imaging: the RUN DMC study. J Alzheimers Dis 2012;32:667-676. 149. van der Flier WM, van Straaten EC, Barkhof F, et al. Small vessel disease and general cognitive function in nondisabled elderly: the LADIS study. Stroke 2005;36:2116-2120.

160 APPENDICES.

150. Van Elderen SS, Zhang Q, Sigurdsson S, et al. Brain Volume as an Integrated Marker for the Risk of Death in a Community-Based Sample: Age Gene/Environment Susceptibility-Reykjavik Study. J Gerontol A Biol Sci Med Sci 2014;30. 1 151. Villarejo A, Benito-Leon J, Trincado R, et al. Dementia-associated mortality at thirteen years in the NEDICES Cohort Study. J Alzheimers Dis 2011;26:543-551. 152. Rothman KJGS, Lash TL. Modern epidemiology, 3rd edition ed: Philadelphia: Wolters Kluwer Health/Lippincott Williams & Wilkins, 2008. 2 153. Wahlund LO, Barkhof F, Fazekas F, et al. A new rating scale for age-related white matter changes applicable to MRI and CT. Stroke 2001;32:1318-1322. 154. Jensen JH, Helpern JA. Effect of gradient pulse duration on MRI estimation of the diffusional kurtosis for a two-compartment exchange model. J Magn Reson 2011;210:233-237. 3 155. Ghafoorian M, Karssemeijer N, Heskes T, et al. Small white matter lesion detection in cerebral small vessel disease. Proceedings of the SPIE Medical Imaging 2015. 156. Pantoni L, Basile AM, Pracucci G, et al. Impact of age-related cerebral white matter changes on the transition to disability -- the LADIS study: rationale, design and methodology. Neuroepidemiology 4 2005;24:51-62. 157. Zheng JJ, Delbaere K, Close JC, Sachdev PS, Lord SR. Impact of white matter lesions on physical functioning and fall risk in older people: a systematic review. Stroke 2011;42:2086-2090. 158. Stern Y. Cognitive reserve. Neuropsychologia 2009;47:2015-2028. 5 159. Brickman AM, Siedlecki KL, Muraskin J, et al. White matter hyperintensities and cognition: testing the reserve hypothesis. Neurobiol Aging 2011;32:1588-1598. 160. Elbaz A, Vicente-Vytopilova P, Tavernier B, et al. Motor function in the elderly: evidence for the reserve hypothesis. Neurology 2013;81:417-426. 6 161. Bohannon RW. Comfortable and maximum walking speed of adults aged 20-79 years: reference values and determinants. Age Ageing 1997;26:15-19. 162. Prins ND, Scheltens P. White matter hyperintensities, cognitive impairment and dementia: an update. Nat Rev Neurol 2015;11:157-165. 7 163. Callisaya ML, Blizzard CL, Wood AG, Thrift AG, Wardill T, Srikanth VK. Longitudinal Relationships Between Cognitive Decline and Gait Slowing: The Tasmanian Study of Cognition and Gait. J Gerontol A Biol Sci Med Sci 2015. 8 164. Bolandzadeh N, Liu-Ambrose T, Aizenstein H, et al. Pathways linking regional hyperintensities in the brain and slower gait. Neuroimage 2014;99:7-13. 165. de Laat KF, van Norden AG, Gons RA, et al. Cerebral white matter lesions and lacunar infarcts contribute to the presence of mild parkinsonian signs. Stroke 2012;43:2574-2579. 9 166. Hatate J, Miwa K, Matsumoto M, et al. Association between cerebral small vessel diseases and mild parkinsonian signs in the elderly with vascular risk factors. Parkinsonism Relat Disord 2016;26:29-34. 167. Bohnen NI, Muller ML, Zarzhevsky N, et al. Leucoaraiosis, nigrostriatal denervation and motor symptoms in Parkinson’s disease. Brain 2011;134:2358-2365. A 168. Ragno M, Berbellini A, Cacchio G, et al. Parkinsonism is a late, not rare, feature of CADASIL: a study on Italian patients carrying the R1006C mutation. Stroke 2013;44:1147-1149.

161 PART VI

169. Korczyn AD. Vascular parkinsonism-characteristics, pathogenesis and treatment. Nat Rev Neurol 2015;11:319-326. 170. Moretti R, Cavressi M, Tomietto P. Gait and apathy as relevant symptoms of subcortical vascular dementia. Am J Alzheimers Dis Other Demen 2015;30:390-399. 171. Vesely B, Rektor I. The contribution of white matter lesions (WML) to Parkinson’s disease cognitive impairment symptoms: A critical review of the literature. Parkinsonism Relat Disord 2016;22 Suppl 1:S166-170. 172. Reijmer YD, Fotiadis P, Martinez-Ramirez S, et al. Structural network alterations and neurological dysfunction in cerebral amyloid angiopathy. Brain 2015;138:179-188. 173. Holtmannspotter M, Peters N, Opherk C, et al. Diffusion magnetic resonance histograms as a surrogate marker and predictor of disease progression in CADASIL: a two-year follow-up study. Stroke 2005;36:2559-2565. 174. Verlinden VJ, van der Geest JN, de Groot M, et al. Structural and microstructural brain changes predict impairment in daily functioning. Am J Med 2014;127:1089-1096.e1082. 175. Jokinen H, Schmidt R, Ropele S, et al. Diffusion changes predict cognitive and functional outcome: the LADIS study. Ann Neurol 2013;73:576-583. 176. Sedaghat S, Cremers LG, de Groot M, et al. Lower microstructural integrity of brain white matter is related to higher mortality. Neurology 2016;87:927-934. 177. Chabriat H, Herve D, Duering M, et al. Predictors of Clinical Worsening in Cerebral Autosomal Dominant Arteriopathy With Subcortical Infarcts and Leukoencephalopathy: Prospective Cohort Study. Stroke 2016;47:4-11. 178. Benjamin P, Zeestraten E, Lambert C, et al. Progression of MRI markers in cerebral small vessel disease: sample size considerations for clinical trials. J Cereb Blood Flow Metab 2015. 179. Bath PM, Wardlaw JM. Pharmacological treatment and prevention of cerebral small vessel disease: a review of potential interventions. Int J Stroke 2015;10:469-478. 180. Godin O, Tzourio C, Maillard P, Mazoyer B, Dufouil C. Antihypertensive treatment and change in blood pressure are associated with the progression of white matter lesion volumes: the Three-City (3C)-Dijon Magnetic Resonance Imaging Study. Circulation 2011;123:266-273. 181. Dufouil C, Chalmers J, Coskun O, et al. Effects of blood pressure lowering on cerebral white matter hyperintensities in patients with stroke: the PROGRESS (Perindopril Protection Against Recurrent Stroke Study) Magnetic Resonance Imaging Substudy. Circulation 2005;112:1644-1650. 182. Doi T, Makizako H, Shimada H, et al. Objectively measured physical activity, brain atrophy, and white matter lesions in older adults with mild cognitive impairment. Exp Gerontol 2015;62:1-6. 183. Benedict C, Brooks SJ, Kullberg J, et al. Association between physical activity and brain health in older adults. Neurobiol Aging 2013;34:83-90. 184. Moll van Charante EP, Richard E, Eurelings LS, et al. Effectiveness of a 6-year multidomain vascular care intervention to prevent dementia (preDIVA): a cluster-randomised controlled trial. Lancet 2016;388:797-805.

162 APPENDICES.

185. Ngandu T, Lehtisalo J, Solomon A, et al. A 2 year multidomain intervention of diet, exercise, cognitive training, and vascular risk monitoring versus control to prevent cognitive decline in at-risk elderly people (FINGER): a randomised controlled trial. Lancet 2015;385:2255-2263. 1 186. Sudlow CL, Warlow CP. Comparable studies of the incidence of stroke and its pathological types: results from an international collaboration. International Stroke Incidence Collaboration. Stroke 1997;28:491-499. 187. Gorelick PB, Scuteri A, Black SE, et al. Vascular contributions to cognitive impairment and 2 dementia: a statement for healthcare professionals from the american heart association/american stroke association. Stroke 2011;42:2672-2713. 188. Benjamin P, Viessmann O, MacKinnon AD, Jezzard P, Markus HS. 7 Tesla MRI in cerebral small vessel disease. Int J Stroke 2015;10:659-664. 3 189. Wardlaw JM, Smith C, Dichgans M. Mechanisms of sporadic cerebral small vessel disease: insights from neuroimaging. Lancet Neurol 2013;12:483-497.

163

A3 Acknowledgements | Dankwoord PART VI

166 APPENDICES.

Dit proefschrift zou niet tot stand gekomen zijn zonder de medewerking, inspiratie, steun en bijdrage van vele personen. Een aantal wil ik graag in het bijzonder bedanken. 1 Allereerst wil ik alle deelnemers aan de RUN DMC studie bedanken voor hun enthousiaste en belangeloze deelname aan ons onderzoek. Door uw inzet konden we uitgebreide informatie verzamelen die de studies beschreven in dit proefschrift mogelijk hebben gemaakt. Geweldig om te merken dat u ruim 10 jaar na de start van de studie nog steeds zo betrokken bent bij 2 het onderzoek.

Mijn bijzondere dank gaat uit naar mijn promotor Prof. dr. F.E. de Leeuw en mijn copromotor Dr. A.M. Tuladhar. 3

Beste Frank-Erik, als geneeskunde student enthousiasmeerde jij mij al voor dit mooie en grote onderzoeksproject, de RUN DMC studie. Jouw liefde voor het onderzoek en de vasculaire neurologie zijn aanstekelijk. Dank voor de fijne samenwerking de afgelopen 8 4 jaar en dank voor je vertrouwen in mij. Je enthousiasme, gedrevenheid en kritische blik heb ik zeer gewaardeerd. Bewonderenswaardig hoe snel jij mijn stukken van commentaar en suggesties voorzag, ondanks de vele promovendi die jij inmiddels begeleidt. Mede daardoor heb ik dit proefschrift af gekregen voor de afronding van mijn opleiding. Veel dank daarvoor. 5

Beste Anil, zonder jouw hulp zou dit proefschrift er niet hebben gelegen. Jouw kennis en kunde op imaging gebied zijn van onschatbare waarde geweest. Ik heb ontzettend veel van je geleerd. Dank voor al je hulp, je positivisme, je interesse en persoonlijke begeleiding. 6 Bedankt dat je deur ook altijd open stond voor snel een vraag tussendoor of een kop koffie samen. Je hebt een gave om complexe dingen ineens veel minder complex te laten lijken. Ik ben er trots op dat jij mijn copromotor bent. 7 Prof. dr. C.J.M. Klijn. Beste Karin, in de laatste fase van mijn promotietraject verruilde jij Utrecht voor Nijmegen en raakte je betrokken bij het Nijmeegse vasculaire onderzoek. Dank voor je interesse en betrokkenheid bij mijn onderzoek. Het is geweldig om te zien met hoeveel 8 ambitie, passie en gedrevenheid je leiding geeft aan de afdeling neurologie.

Dr. E.J. van Dijk. Beste Ewoud, er lijken geen grenzen te zitten aan jouw kennis op zowel het gebied van de neurologie als op het doen van onderzoek. Ik heb op beide gebieden 9 ontzettend veel van je geleerd. Dank voor de waardevolle discussies tijdens de vasculaire meetings en je belangstelling en betrokkenheid bij mijn onderzoek.

167 PART VI

Dr. R.A.J. Esselink. Beste Rianne, dank voor je hulp met het parkinsonisme stuk en het screenen van alle deelnemers met hypokinetisch-rigide verschijnselen voor onze studie. Wat een extra werk heb jij verzet. Jouw kennis en kunde op het gebied van de bewegingsstoornissen en de discussies met jou over dit manuscript zijn zeer waardevol geweest. Ook veel dank voor je persoonlijk interesse en de fijne samenwerking.

Prof. dr. Padberg, Dr. Kappelle en Dr. Post, mijn opleiders, dank voor de ruimte en tijd die ik tijdens mijn opleiding heb gekregen voor mijn promotietraject. Bart Post, beste Bart, ik bewonder je gedrevenheid, bevlogenheid en passie voor het vak en het opleiden. Dank voor je persoonlijke belangstelling en de fijne opleiding de afgelopen jaren.

Dank aan de manuscript commissie, Prof. dr. ir. N. Karssemijer, Dr. V.G.M. Weerdesteyn en Prof. Dr. W.M. van der Flier, voor het beoordelen van mijn proefschrift.

Een speciaal woord van dank voor Inge van Uden. Lieve Inge, mijn onderzoeksmaatje, zonder jou was dit vervolgonderzoek van de RUN DMC studie niet zo’n succes geworden. Het terugzien van bijna 400 deelnemers was een hele klus, die zoveel leuker werd door het samen met jou te doen. Dank voor je geweldige organisatietalent, je oplossingsgerichtheid, je humor en aanstekelijke lach. We vulden elkaar goed aan en hebben alle onderzoeksperikelen samen overwonnen. Nu hebben we allebei het boekje af, de kroon op ons werk!

Het RUN DMC onderzoeksteam. Karlijn de Laat, Anouk van Norden en Rob Gons, de eerste generatie en grondleggers van de RUN DMC studie, dankzij jullie werk en inspanningen konden we een vliegende start maken met het eerste vervolg van de RUN DMC studie. Dank hiervoor en dank voor jullie blijvende betrokkenheid en het zijn van een vraagbaak. De derde en vierde generatie RUN DMC-ers: Mayra Bergkamp, Esther van Leijsen, Annemiek ter Telgte en Kim Wiegertjes. Geweldig om te zien met hoeveel enthousiasme en succes jullie de studie voortzetten. Dank voor de leuke congresbezoeken samen, de nuttige discussies over het onderzoek en de gezelligheid op de onderzoekskamer op de 5e. Ik wens jullie veel succes met jullie onderzoek en verdere carrière.

Een speciaal woord van dank aan alle mede-auteurs voor de waardevolle bijdrage aan de manuscripten in dit proefschrift. Loes, bedankt voor je hulp bij het maken van de predicitiemodellen en je geduld met mijn beperkte statistische kennis. Moshen Ghafoorian, thank you for developing a semi-automatic white matter hyperintensity segmentation method, which was of great help to our delta imaging manuscript on gait decline.

168 APPENDICES.

Natuurlijk ook veel dank aan de rest van het vasculaire (onderzoeks)team (Renate, Noortje, Mayte, Pauline, Nathalie, Frank, Edo, Tessa, Merel, Bram, Karin, Sharon, Saskia en Annet). Bedankt voor de gezelligheid, fijne samenwerking en waardevolle input tijdens 1 onderzoeksbesprekingen. Het is geweldig om deel uit te mogen maken van zo’n getalenteerd vasculair team.

Mijn dank gaat ook uit naar iedereen die onze data verzameling heeft ondersteund of erbij 2 heeft geholpen. Dank aan de assistentes van de poli neurologie voor hun hulp, maar ook geduld als ik de gang weer eens bezette met de looptesten en voor het horen van iedere keer dezelfde uitleg hierbij, inclusief ‘1-2-3-start’! Dank aan Michel Verbruggen, voor de altijd snelle hulp en ondersteuning indien er zich 3 technische problemen voordeden met de GAITRite. Tevens een woord van dank aan de secretaresses van het Donders centrum voor hun ondersteuning en het opvangen van onze proefpersonen. In het bijzonder ook aan Paul Gaalman. Paul, bedankt dat je ons wegwijs hebt gemaakt in de wereld van het scannen en 4 voor het gezellig meevieren van onze mijlpalen hierin, inclusief slingers en taart. Tenslotte wil ik graag een aantal studenten bedanken voor hun hulp en bijdrage aan het RUN DMC onderzoek: Heleen van den Berg, Willemijn Geense, Valerie Lohner en Inge van der Holst. Veel dank voor jullie inzet. 5

Mijn collega’s van de neurologie. Wat heb ik genoten van mijn opleiding en wat was het fijn om in zo’n goed team te werken. Ik ga jullie, en de assistentenweekenden en skireisjes zeker missen! 6 Een speciaal woord van dank aan mijn ‘jaargenoten’ Judith van Gaalen, Nicolien van der Kolk, Renate Arntz en Inge van Uden. Wat heb ik ontzettend geboft met het feit dat we snel na elkaar startten met de opleiding en er al snel een fijne vriendschap ontstond. Ik heb ongelofelijk veel plezier met jullie gehad. Samen onderzoek doen op 1 kamer was soms net 7 iets te gezellig ;-). Ik ga het missen dat ik jullie niet meer bijna dagelijks zie. Hopelijk zetten we onze gezellige etentjes, borrels, sinterklaasavonden en congresbezoeken nog heel lang voort. 8 Joyce van der Vegt. Lieve Joyce, van mentormama tijdens de geneeskunde introductie, naar vriendin en collega. Dank voor de fijne vriendschap, je steun en de gezellige etentjes, waarbij ik vaak mocht genieten van je geweldige kookkunst. 9

Mijn vrienden van de studietijd. Lieve Marjolein (E) en Mijke, dank voor jullie fijne vriendschap en de leuke uitjes. Ook al zien we elkaar weinig, onze afspraakjes zijn altijd des te gezellig! Jullie weten als geen ander wat er allemaal bij promoveren komt kijken. Dank voor jullie A hulp, steun en interesse.

169 PART VI

Lieve skiclub vrienden (Jacolien, Leon, Maaike, Marcel, Marieke, Bruno, Marjolein (A) en Marvin), ook al hebben we al een tijdje niet meer geskied samen, de ‘après-ski’ afspraakjes zijn altijd erg gezellig. Dank voor jullie vriendschap en betrokkenheid. Hopelijk gaan we in de toekomst weer skiën samen. Lieve Anneke, Erica, Jacolien, Katrien, Marit, Sally en Suzan, wat ben ik bevoorrecht met vriendinnen zoals jullie. De passie voor muziek bracht ons samen in onze studententijd. Inmiddels delen we lief en leed. Dank voor jullie interesse en niet aflatende steun bij het afronden van dit proefschrift. Er zit nog altijd veel muziek in onze vriendschap, jullie zijn geweldig. Op naar onze volgende Qladies date! Lieve Daphne en Maaike, mijn studievriendinnen van het eerste uur, wat bof ik met zulke lieve, trouwe en hartelijke vriendinnen. Met jullie was mijn studententijd zoveel leuker! Dank voor al jullie steun, belangstelling en gezelligheid en dat jullie er altijd voor me zijn. Lieve Daphne, wat een geluk dat we allebei in Nijmegen konden blijven voor onze specialisatie en ook nog in dezelfde periode konden starten met een promotieonderzoek. Het was ontzettend fijn om samen te praten over de ups en downs van onderzoek doen en om onze successen samen te vieren. Dank dat ik bij jou altijd welkom ben. Ik voel me vereerd dat jij mijn paranimf wil zijn.

Lieve Marieke, Janneke, Inge en Geert, wat ben ik blij met jullie als zussen en broer. In onze jeugd waren we al een hecht team en deden we graag veel samen. Ondanks dat we elkaar nu minder zien, voel ik dit nog steeds zo. Ik koester onze gezellige theedrinkmomenten, etentjes, uitjes en familieweekenden. Dank dat ik bij jullie altijd thuis ben. Ik ben trots op jullie! Lieve Marieke en Janneke, het zijn van drielingzussen en het alle drie doorlopen van een medische studie schept een zeer bijzondere band. Jullie begrijpen mij als geen ander. En zoals met alles, hebben jullie me ook bij mijn promotieonderzoek van uitgebreid advies en steun voorzien. Veel dank. Lieve Janneke, ik vind het een eer dat jij mijn paranimf wil zijn en vind het erg fijn dat jij op deze bijzondere dag ook letterlijk naast me staat. Lieve Inge, dank voor je hulp bij het onderzoek. Mede door jouw inzet hadden we de gegevens van ons onderzoek veel sneller compleet. Extra leuk dat ik met jou ook inhoudelijk kon discussiëren over het onderzoek en je me van de nodige epidemiologische en statistische kennis voorzag. Ik heb het ook erg gewaardeerd dat je vaak zorgde voor de nodige muzikale afleiding. Hopelijk spelen we nog heel lang muziek samen. Lieve Geert, mijn stoere broer(tje ;-)), door je drukke werk op onregelmatige tijden zien wij elkaar helaas weinig. Dank voor je humor en je vaak grappige berichtjes over de app.

Lieve Tom, Olivier en Bas, ik kan me geen betere zwagers wensen. Dank dat jullie deur altijd voor me open staat en jullie bereidheid me altijd te helpen.

170 APPENDICES.

Lieve Mees, Nina en Fien, wat een geweldig neefje en lieve nichtjes zijn jullie. Met jullie betoverende lachjes en enthousiaste verhaaltjes doen jullie mij alles om me heen even helemaal vergeten. 1

Lieve papa en mama, ik kan niet in woorden uit drukken hoe dankbaar ik ben en hoe belangrijk jullie voor me zijn, daarom is dit boekje voor jullie. Dank voor jullie onvoorwaardelijke liefde en steun en dank voor alle kansen die jullie me hebben geboden. Mama, veel dank voor je 2 luisterend oor en onze gezellige uitjes en (mini)vakanties op de momenten dat ik het juist zo nodig had. Papa, veel dank voor je bijdrage aan dit boekje. Ik vind het ontzettend bijzonder dat jij de voorkant van dit boekje hebt gefotografeerd. Ik ben trots op het resultaat; het had niet mooier gekund. 3

171

A4 Curriculum vitae PART VI

174 APPENDICES.

Ellen van der Holst was born on July 31th 1984 in Nijmegen as one of triplets. She attended secondary school at the 1 Marianum in Groenlo and graduated in 2002. That year she started medical school at the Radboud University Nijmegen. During this period she followed the 2 extracurricular ‘Honours Program’ of the Radboud University (2004-2006) and performed a clinical elective on tropical medicine in Mangochi District Hospital in 3 Malawi (2006). In 2008, she performed a research internship under the supervision of Prof. dr. FE de Leeuw, entitled ‘Diffusion tensor imaging of the cingulum and memory performance’. In November 2008, she obtained her medical degree and she started working as a resident at the department of Neurology, Radboud University Medical Centre, Nijmegen. In June 4 2009, she started her specialization (Prof. dr. G.W.A.M. Padberg (2009-2014), Dr. A.C. Kappelle (2014-2015) and Dr. B. Post (2015-2017)). During her specialization she was a member of the board of the association of residents of the Radboud University Medical Centre (AAVR) (2010-2015), as well as a board member of the association of neurology residents of the 5 Radboud University Medical Centre (2015-2016). In 2011, she started her PhD project, under the supervision of Prof. dr. FE de Leeuw, which resulted in this thesis. In 2013, she won the ‘young investigators award’ of the European Stroke Conference, London, United Kingdom. She finished her specialization in January 2017 and as of February 2017, she is 6 working as a neurologist at the Jeroen Bosch Hospital in ‘s-Hertogenbosch.

175

A5 List of publications PART VI

178 APPENDICES.

Van der Holst HM, van Uden IWM, de Laat KF, van Leijsen EMC, van Norden AGW, Norris DG, van Dijk EJ, Tuladhar AM, de Leeuw FE. Baseline cerebral small vessel disease is not associated with gait decline after 5 years. Movement Disorders Clinical Practice, 2016 Nov. 1 DOI:10.1002/mdc3.12435

Van Uden IWM, van der Holst HM, van Leijsen EMC, Tuladhar AM, van Norden AGW, de Laat KF, Claassen JA, van Dijk EJ, Kessels RP, Richard E, Tendolkar I, de Leeuw FE. Late-onset 2 depressive symptoms increase the risk of dementia in small vessel disease. Neurology, 2016 Sept; 87(11):1102-9.

Van der Holst HM, van Uden IWM, Tuladhar AM, de Laat KF, van Norden AGW, Norris DG, 3 van Dijk EJ, Rutten-Jacobs LC, de Leeuw FE. Factors associated with 8-year mortality in older patients with cerebral small vessel disease. The RUN DMC Study. JAMA Neurology, 2016 Apr; 73(4):402-9. 4 Tuladhar AM, van Uden IWM, Rutten-Jacobs LC, Lawrence A, van der Holst HM, van Norden AGW, de Laat KF, van Dijk EJ, Claassen JA, Kessels RP, Markus HS, Norris DG, de Leeuw FE. Structural network efficiency predicts conversion to dementia. Neurology, 2016 Mar; 86(12):1112-9. 5 van Uden IWM*, Tuladhar AM*, van der Holst HM, van Leijsen EMC, van Norden AGW, de Laat KF, Rutten-Jacobs LC, Norris DG, Claassen JA, van Dijk EJ, Kessels RP, de Leeuw FE. Diffusion tensor imaging of the hippocampus predicts the risk of dementia; the RUN DMC study. Hum 6 Brain Mapp. 2016 Jan;37(1):327-37. van Uden IWM, van der Holst HM, Tuladhar AM, van Norden AGW, de Laat KF, Rutten- Jacobs LC, Norris DG, Claassen JA, van Dijk EJ, Kessels RP, de Leeuw FE. White Matter and 7 Hippocampal Volume Predict the Risk of Dementia in Patients with Cerebral Small Vessel Disease: The RUN DMC Study. J Alzheimers Dis. 2015 Nov;49(3):863-73.

8 van der Holst HM, van Uden IWM, Tuladhar AM, de Laat KF, van Norden AGW, Norris DG, van Dijk EJ, Esselink RAJ, Platel B, de Leeuw FE. Cerebral small vessel disease and incident parkinsonism: The RUN DMC study. Neurology, 2015 Nov;85(18):1569-77. 9 van Uden IWM, van der Holst HM, Schaapsmeerders P, Tuladhar AM, van Norden AGW, de Laat KF, Norris DG, Claassen JA, van Dijk EJ, Richard E, Kessels RP, de Leeuw FE. Baseline white matter microstructural integrity is not related to cognitive decline after 5 years: The RUN DMC study. BBA Clin., 2015 Oct;4:108-114. A

179 PART VI

van der Holst HM*, Tuladhar AM*, van Norden AGW, de Laat KF, van Uden IWM, van Oudheusden LJ, Zwiers MP, Norris DG, Kessels RP, de Leeuw FE. Microstructural integrity of the cingulum is related to verbal memory performance in elderly with cerebral small vessel disease: the RUN DMC study. Neuroimage, 2013 Jan;65:416-23.

Submitted

Van der Holst HM*, Tuladhar AM*, Zerbi V, van Uden IWM, de Laat KF, van Leijsen EMC, Ghafoorian M, Platel B, Bergkamp MI, van Norden AG, Norris DG, van Dijk EJ, Kiliaan AJ, de Leeuw FE. White matter atrophy and loss of white matter integrity are associated with gait decline in cerebral small vessel disease.

Bergkamp MI, Tuladhar AM, van der Holst HM, van Leijsen EMC, Ghafoorian M, van Uden IWM, van Dijk EJ, Norris DG, Platel B, Esselink RAJ, de Leeuw FE. Brain atrophy increases risk of parkinsonism in cerebral small vessel disease.

Van Uden IWM*, van Leijsen EMC*, Ghafoorian M, Bergkamp MI, Lohner V, Kooijmans ECM, van der Holst HM, Tuladhar AM, Norris DG, van Dijk EJ, Rutten-Jacobs LCA, Platel B, Klijn CJM, de Leeuw FE. The rise and fall of cerebral small vessel disease: the RUN DMC study.

*Shared first authorship

180 APPENDICES.

181

A6 Dissertations of the disorders of movement research group, Nijmegen PART VI

184 APPENDICES.

Vascular disorders of movement – The Radboud Stroke centre • Liselore Snaphaan. Epidemiology of post stroke behavioral consequences. Radboud University Nijmegen, 12 March 2010 1 • Karlijn F. de Laat. Motor performance in individuals with cerebral small vessel disease: an MRI study. Radboud University Nijmegen, 29 November 2011 • Anouk G.W. van Norden. Cognitive function in elderly individuals with cerebral small vessel disease. An MRI study. Radboud University Nijmegen, 30 November 2 2 • Rob Gons. Vascular risk factors in cerebral small vessel disease. A diffusion tensor imaging study. Radboud University Nijmegen, 10 December 2012 • Loes C.A. Rutten-Jacobs. Long-term prognosis after stroke in young adults. Radboud University Nijmegen, 14 April 2014 3 • Noortje A.M.M. Maaijwee. Long-term neuropsychological and social consequences after stroke in young adults. Radboud University Nijmegen, 12 June 2015 • Nathalie E. Synhaeve. Determinants of long-term functional prognosis after stroke in young adults. Radboud University Nijmegen, 28 September 2016 4 • Anil M. Tuladhar. The disconnected brain: mechanisms of clinical symptoms in small vessel disease. Radboud University Nijmegen, 4 October 2016 • Pauline Schaapsmeerders. Long-term cognitive impairment after first-ever ischemic stroke in Young adults: a neuroimaging study. Radboud University Nijmegen, 5 24 January 2016 • Ingeborg W.M. van Uden. Behavioural consequences of cerebral small vessel disease; an MRI approach. Radboud University Nijmegen, 14 February 2017 • Renate M. Arntz. The long-term risk of vascular disease and epilepsy after stroke in 6 young adults. Radboud University Nijmegen, 16 February 2017

Parkinson Centre Nijmegen (ParC) • Jasper E. Visser. The basal ganglia and postural control. Radboud University 7 Nijmegen, 17 June 2008 • Maaike Bakker. Supraspinal control of walking: lessons from motor imagery. Radboud University Nijmegen, 27 May 2009 8 • W. Farid Abdo. Parkinsonism: possible solutions to a diagnostic challenge. Radboud University Nijmegen, 7 October 2009 • Samyra H.J. Keus. Physiotherapy in Parkinson’s disease. Towards evidence-based practice. Leiden University, 29 April 2010 9 • Lars B. Oude Nijhuis. Modulation of human balance reactions. Radboud University Nijmegen, 29 November 2010 • Maarten J. Nijkrake. Improving the quality of allied health care in Parkinson’s disease through community-based networks: the ParkinsonNet health care concept. A Radboud University Nijmegen, 29 November 2010

185 PART VI

• Rick C.G. Helmich. Cerebral reorganization in Parkinson’s disease. Radboud University Nijmegen, 24 May 2011 • Charlotte A. Haaxma. New perspectives on preclinical and early stage Parkinson’s disease. Radboud University Nijmegen, 6 December 2011 • Johanna G. Kalf. Drooling and dysphagia in Parkinson’s disease. Radboud University Nijmegen, 22 December 2011 • Anke H. Snijders. Tackling freezing of gait in Parkinson’s disease. Radboud University Nijmegen,4 June 2012 • Bart F.L. van Nuenen. Cerebral reorganization in premotor parkinsonism. Radboud University Nijmegen, 22 November 2012 • Wandana Nanhoe-Mahabier. Freezing of physical activity in Parkinson’s disease, the challenge to change behaviour. Radboud University Nijmegen, 13 February 2013 • Marlies van Nimwegen. Promotion of physical activity in Parkinson’s disease, the challenge to change behaviour. Radboud University Nijmegen, 6 March 2013 • Arlène D. Speelman. Promotion of physical activity in Parkinson’s disease, feasibility and effectiveness. Radboud University Nijmegen, 6 March 2013 • Tjitske Boonstra. The contribution of each leg to bipedal balance control. University Twente, 6 June 2013 • Marjolein A van der Marck. The Many faces of Parkinson’s disease: towards a multifaceted approach? Radboud University Nijmegen, 10 January 2014 • Katrijn Smulders. Cognitive control of gait and balance in patients with chronic stroke and Parkinson’s disease. Radboud University Nijmegen, 21 May 2014 • Marjolein B. Aerts. Improving diagnostic accuracy in parkinsonism. Radboud University Nijmegen, 27 June 2014 • Maartje Louter. Sleep in Parkinson’s disease. A focus on nocturnal movements. Radboud University Nijmegen, 13 February 2015 • Frederick Anton Meijer. Clinical Application of Brain MRI in Parkinsonism: From Basic to Advanced Imaging, Radboud University Nijmegen, 23 June 2015 • Jorik Nonnekes. Balance and gait in neurodegenerative disease: what startle tells us about motor control, Radboud University Nijmegen, 2 September 2015 • Martijn van der Eijk. Patient-centered care in Parkinson’s disease. Radboud University Nijmegen, 1 December 2015 • Ingrid Sturkenboom. Occupational therapy for people with Parkinson’s disease: towards evidence-informed care. Radboud University Nijmegen, 11 February 2016 • Merel M. van Gilst. Sleep benefit in Parkinson’s disease. Radboud University Nijmegen, 13 April 2016 • Arno M. Janssen. Transcranial magnetic stimulation - measuring and modelling in health and disease. Radboud University Nijmegen, 2 June 2016

186 APPENDICES.

Non-Parkinsonian disorders of movement • Sacha Vermeer. Clinical and genetic characterization of autosomal recessive cerebellar ataxias. Radboud University Nijmegen, 5 April 2012 1 • Susanne T. de Bot. Hereditary spastic paraplegias in the Netherlands. Radboud University Nijmegen, 20 December 2013 • Catherine C.S. Delnooz. Unravelling primary focal dystonia. A treatment update and new pathophysiological insights. Radboud University Nijmegen, 7 January 2014 2 • Ella M.R. Fonteyn. Falls, physiotherapy, and training in patients with degenerative ataxias. Radboud University Nijmegen, 29 June 2016.

Neuromuscular disorders of movement 3 • Mireille van Beekvelt. Quantitative near infrared spectroscopy (NIRS) in human skeletal muscle. Radboud University Nijmegen, 24 April 2002 • Johan Hiel. Ataxia telangiectasia and Nijmegen Breakage syndrome, neurological, immunological and genetic aspects. Radboud University Nijmegen, 23 April 2004 4 • Gerald JD Hengstman. Myositis specific autoantibodies, specificity and clinical applications. Radboud University Nijmegen, 21 September 2005 • M. Schillings. Fatigue in neuromuscular disorders and chronic fatigue syndrome, a neurophysiological approach. Radboud University Nijmegen, 23 November 2005 5 • Bert de Swart. Speech therapy in patients with neuromuscular disorders and Parkinson’s disease. Diagnosis and treatment of dysarthria and dysphagia. Radboud University Nijmegen, 24 march 2006 • J. Kalkman. From prevalence to predictors of fatigue in neuromuscular disorders. 6 The building of a model. Radboud University Nijmegen, 31 October 2006 • Nens van Alfen. Neuralgic amyotrophy. Radboud University Nijmegen, 1 November 2006 • Gea Drost. High-density surface EMG, pathophysiological insights and clinical 7 applications. Radboud University Nijmegen, 9 March 2007 • Maria Helena van der Linden. Pertubations of gait and balance: a new experimental setup applied to patients with CMT type 1a. Radboud University Nijmegen, 6 October 8 2009 • Jeroen Trip. Redefining the non-dystrophic myotonic syndromes. Radboud University Nijmegen, 22 January 2010 • Corinne G.C. Horlings. A weak balance: balance and falls in patients with 9 neuromuscular disorders. Radboud University Nijmegen, 1 April 2010 • E. Cup. Occupational therapy, physical therapy and speech therapy for persons with neuromuscular diseases, an evidence based orientation. Radboud University Nijmegen, 5 July 2011 A • Alide Tieleman. Myotonic dystrophy type 2, a newly diagnosed disease in the Netherlands. Radboud University Nijmegen, 15 July 2011

187 PART VI

• Nicol Voermans. Neuromuscular features of Ehlers-Danlos syndrome and Marfan syndrome. Radboud University Nijmegen, 2 September 2011 • Allan Pieterse. Referral and indication for occupational therapy, physical therapy and speech- language therapy for persons with neuromuscular disorders. Radboud University Nijmegen, 13 February 2012 • Bart Smits. Chronic Progressive External Ophthalmoplegia more than meets the eye. Radboud University Nijmegen, 5 June 2012 • Ilse Arts. Muscle ultrasonography in ALS. Radboud University Nijmegen, 31 October 2012 • M. Minis. Sustainability of work for persons with neuromuscular diseases. Radboud University Nijmegen, 13 November 2013 • Willemijn Leen. Glucose transporter – 1 deficiency syndrome. Radboud University Nijmegen, 26 June 2014 • Barbara Jansen. Magnetic Resonance Imaging signature of fascioscapulohumeral muscular dystrophy. Radboud University Nijmegen, 14 September 2015 • Noortje Rijken. Balance and gait in FSHD, relations with individual muscle involvement. Radboud University Nijmegen, 8 December 2015 • Femke Seesing. Shared Medical appointments for neuromuscular patients and their partners. Radboud University Nijmegen, 2 September 2016 • Nicole Voet. Aerobic exercise and cognitive behavioral therapy in fascioscapulohumeral dystrophy: a model based approach. Radboud University Nijmegen, 14 October 2016.

188 APPENDICES.

189

A7 Donders Graduate School for Cognitive Neuroscience Series PART VI

192 APPENDICES.

For a successful research Institute, it is vital to train the next generation of young scientists. To achieve this goal, the Donders Institute for Brain, Cognition and Behaviour established the Donders Graduate School for Cognitive Neuroscience (DGCN), which was officially 1 recognised as a national graduate school in 2009. The Graduate School covers training at both Master’s and PhD level and provides an excellent educational context fully aligned with the research programme of the Donders Institute. The school successfully attracts highly talented national and international students in 2 biology, physics, psycholinguistics, psychology, behavioral science, medicine and related disciplines. Selective admission and assessment centres guarantee the enrolment of the best and most motivated students. The DGCN tracks the career of PhD graduates carefully. More than 50% of PhD alumni show 3 a continuation in academia with postdoc positions at top institutes worldwide, e.g. Stanford University, University of Oxford, University of Cambridge, UCL London, MPI Leipzig, Hanyang University in South Korea, NTNU Norway, University of Illinois, North Western University, Northeastern University in Boston, ETH Zürich, University of Vienna etc.. Positions outside 4 academia spread among the following sectors: specialists in a medical environment, mainly in genetics, geriatrics, psychiatry and neurology. Specialists in a psychological environment, e.g. as specialist in neuropsychology, psychological diagnostics or therapy. Positions in higher education as coordinators or lecturers. A smaller percentage enters 5 business as research consultants, analysts or head of research and development. Fewer graduates stay in a research environment as lab coordinators, technical support or policy advisors. Upcoming possibilities are positions in the IT sector and management position in pharmaceutical industry. In general, the PhDs graduates almost invariably continue with 6 high-quality positions that play an important role in our knowledge economy. For more information on the DGCN as well as past and upcoming defenses please visit: http://www.ru.nl/donders/graduate-school/donders-graduate/ 7

193 Paranimfen Paranimfen Daphne Everaerd 6521 JT Nijmegen Daphne Everaerd 6521 JT Nijmegen Ellen van der Holst Ellen van Janneke van der Holst van Janneke Van den Havestraat 44 Havestraat den Van Ellen van der Holst Ellen van Janneke van der Holst van Janneke Van den Havestraat 44 Havestraat den Van Mind the step Mind the step Mind the step Mind the step aansluitende receptie. receptie. aansluitende deze plechtigheid en de deze [email protected] precies in de Aula van de van in de Aula precies aansluitende receptie. receptie. aansluitende U bent van harte welkom bij bij welkom harte U bent van deze plechtigheid en de deze [email protected] Comeniuslaan 2 te Nijmegen Comeniuslaan 2 te [email protected] precies in de Aula van de van in de Aula precies [email protected] Radboud Universiteit Nijmegen, Nijmegen, Universiteit Radboud U bent van harte welkom bij bij welkom harte U bent van [email protected] verdediging van mijn proefschrift van verdediging Comeniuslaan 2 te Nijmegen Comeniuslaan 2 te UITNODIGING [email protected] Voor het bijwonen van de openbare de openbare bijwonen van het Voor Radboud Universiteit Nijmegen, Nijmegen, Universiteit Radboud Op woensdag 5 april 2017 om 14.30u verdediging van mijn proefschrift van verdediging UITNODIGING Voor het bijwonen van de openbare de openbare bijwonen van het Voor Brain changes in motor performance in motor changes Brain in cerebral small vessel disease disease vessel small in cerebral Op woensdag 5 april 2017 om 14.30u Brain changes in motor performance in motor changes Brain in cerebral small vessel disease disease vessel small in cerebral

Mind the step the step Mind Mind the step the step Mind Ellen (H.M.) van der Holst Ellen (H.M.) van Ellen (H.M.) van der Holst Ellen (H.M.) van Brain changes in motor performance in motor changes Brain in cerebral small vessel disease small vessel in cerebral Brain changes in motor performance in motor changes Brain in cerebral small vessel disease small vessel in cerebral

THEMIND LONG-TERM THE STEP RISK IN CEREBRAL OF VASCULAR SMALL DISEASE VESSEL AND DISEASE EPILEPSY Brain AFTER changes STROKE in INmotor YOUNG performance ADULTS Ellen (H.M.) van der HolstRENATE M ARNTZ 266 THEMIND LONG-TERM THE STEP RISK IN CEREBRAL OF VASCULAR SMALL DISEASE VESSEL AND DISEASE EPILEPSY Brain AFTER changes STROKE in INmotor YOUNG performance ADULTS Ellen (H.M.) van der HolstRENATE M ARNTZ 266 ISBN 978-94-6284-098-0 ISBN 978-94-6284-098-0